DETAILED ACTION
Notice to Applicant
The following is a FINAL Office action upon examination of application number 18/780,671, filed on 07/23/2024. Claims 1-20 are pending in the application and have been examined on the merits discussed below.
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Response to Amendment
In the response filed November 21, 2025, Applicant amended claims 1-5, 10-15, and 19-20, and did not cancel any claims. No new claims were presented for examination.
Applicant's amendments to claims 11-15 and 19 are hereby acknowledged. The amendments are sufficient to overcome the previously issued claim rejection under 35 U.S.C. 112(b); accordingly, this rejection has been withdrawn.
Applicant's amendments to the claims are hereby acknowledged. The amendments are not sufficient to overcome the previously issued claim rejection under 35 U.S.C. 101; accordingly, this rejection has been maintained.
In the previous Office action, allowable subject matter was indicated. Specifically, the Office action stated that claims 4 and 14 were objected to as being dependent upon rejected base claims, but would be allowable if rewritten in independent form including all of the limitations of the base claims and any intervening incorporate the limitations claims. The response did not incorporate the limitations of claim 4 into claim 1 and the limitations of claim 14 into claim 11, thus the pending claims are not in condition for allowance.
Response to Arguments
Applicant's arguments filed November 21, 2025, have been fully considered.
Applicant submits “that the claims do not recite an abstract idea. The independent claims are unrelated to task scheduling, but instead are directed to generating subtasks for a received task. The claim limitations are directed to decomposing a task received from a user into subtasks through application of a generative model to a prompt generated by a task management system, which is distinct from scheduling of tasks. As amended, representative independent claim 1 selects "a set of supplemental tasks based on the list including the task, based on creation dates of tasks associated with the list, and based on measures of similarity between an embedding determined for the task and embeddings determined for tasks associated with the list," applies a generative model to a prompt including at least a title of each of supplemental task, presents candidate subtasks generated by the generative model based at least in part on the title of each supplemental task, and stores a candidate subtask selected by the user in association with the task. This generation of candidate subtasks from a received task and candidate subtasks selected based on criteria specified by independent claim 1 does not schedule a task or schedule one or more candidate subtasks.” [Applicant’s Remarks, 11/21/2025, pages 10-11]
The Examiner respectfully disagrees. The Office action indicated that the claims recite and abstract idea falling within the grouping of certain methods of organizing human activity, specifically managing and organizing tasks. Although Applicant argues that the claims are directed to generating subtasks rather than scheduling tasks, decomposing tasks into subtask is itself a form of task management. The limitations related to receiving a task, selecting related tasks based on dates and similarity measures, generating candidate subtasks, presenting tasks to a user, and storing a selected subtask, collectively concern organizing and managing task related information. For the reasons above, this argument is found unpersuasive.
Applicant submits “Applying a generative model to a prompt including information about a set of supplemental tasks selected by a task management system "based on the list including the task, based on creation dates of tasks associated with the list, and based on measures of similarity between an embedding determined for the task and embeddings determined for tasks associated with the list," and subsequently storing a received selection of a candidate subtask from a user in association with a task is not a fundamental economic principle or practice, is not a commercial or a legal interaction, and is not managing personal behavior or relationships or interactions between people. MPEP § 2106.04(a)(2)(II). Instead, the pending claims identify specific stored information, the set of supplemental tasks, and tunes a generative model to generate a specific output, one or more candidate subtasks, accounting for the supplemental tasks by including the set of supplemental tasks in a prompt for the generative model. Hence, the pending claims are not directed to one of the "certain methods of organizing human activity" recited by MPEP § 2106.04(a)(2)(II).” [Applicant’s Remarks, 11/21/2025, page 11]
The Examiner respectfully disagrees. In response to Applicant’s argument, it is first noted that the Office action did not characterize the claims as reciting a fundamental economic principle or practice, commercial interaction, or legal interaction. Rather, the claims were determined to fall within the “"certain methods of organizing human activity" grouping because they are directed to managing and organizing tasks. Task management constitutes organizing human activity. Moreover, Applicant’s argument that the claims “tune” a generative model, is not supported by the claim language. The claim recites “applying a generative model to a prompt,” but does not recite tuning, modifying, or otherwise altering operation of the generative model itself. For the reasons above, this argument is found unpersuasive.
Applicant submits “Even if the claims were to recite an abstract idea, the pending claims integrate the alleged abstract idea into a practical application.” [Applicant’s Remarks, 11/21/2025, page 11]
In response to Applicant’s argument that “even if the claims were to recite an abstract idea, the pending claims integrate the alleged abstract idea into a practical application,” it is noted that the additional elements in exemplary amended claim 1 are: a task management system, processor, a non-transitory computer readable medium, a generative model, an interface generated by the task management system, which merely serve to tie the abstract idea to a particular technological environment (computer-based operating environment) via generic computing hardware, software/instructions, which is not sufficient to amount to a practical application, as noted in MPEP 2106.05. Applicant has provided no facts/evidence, cited any portion of the Specification, nor provided a persuasive line of reasoning showing how the additional elements are integrated with the abstract idea to integrate the abstract idea into a practical application.
It is also noted that the claims are devoid of any discernible change, transformation, or improvement to a computer (software or hardware) or any existing technology. Applicant has not shown that any specific technological improvement is achieved within the scope of the claims. It bears emphasis that no task management system, processor, generative model, interface, or technological elements are modified or improved upon in any discernible manner. Instead, the result produced by the claims is simply information relating to one or more candidate subtasks, which is not a technical result or improvement thereof.
Furthermore, the additional elements fail to integrate the abstract idea into a practical application because they fail to provide an improvement to the functioning of a computer or to any other technology or technical field, fail to apply the exception with a particular machine, fail to apply the judicial exception to effect a particular treatment or prophylaxis for a disease or medical condition, fail to effect a transformation of a particular article to a different state or thing, and fail to apply/use the abstract idea in a meaningful way beyond generally linking the use of the judicial exception to a particular technological environment. For the reasons above, this argument is found unpersuasive.
Applicant submits “the pending claims present "the one or more candidate subtasks generated based on the task and each supplemental task of the set to the user through an interface generated by the task management system," receive "a selection of a candidate subtask from the user via the interface," and store "the selected candidate subtask in association with the task at the task management system." Hence, the claimed task management system generates candidate subtasks for a task based on application of a generative model to a prompt that includes at least a title of specific supplemental tasks selected "based on a list including a received task from a user based on creation dates of tasks associated with the list, and based on measures of similarity between an embedding determined for the task and embeddings determined for tasks associated with the list" and presents the generated candidate subtasks to the user via an interface. This specifically generates an interface based on the output of the generative model that dynamically generated the candidate subtasks, providing a technological improvement in generation and presentation of candidate subtasks for a task to a user through the task management system.” [Applicant’s Remarks, 11/21/2025, pages 13-14]
The Examiner respectfully disagrees. As best understood by the Examiner, Applicant argues that presenting candidate subtasks through an interface constitutes a technological improvement. However, the claim merely recites displaying information generated by a generative model to a user via an interface and storing user’s selection. Displaying and storing information are routine computer function that do not improve the functioning of the computer itself. The generation of the candidate subtasks using a generative model and their presentation via an interface is directed to implementing the abstract idea of organizing tasks, not to improving computer technology. The interface serves as a medium for outputting information, rather than providing a technical solution to a technical problem. For the reasons above, this argument is found unpersuasive.
Applicant submits “Sodhi's general allusion to "text to be presented to a user" does not disclose or suggest "presenting the one or more candidate subtasks generated based on the task and each supplemental task of the set to the user through an interface generated by the task management system," as recited by amended independent claim 1.” [Applicant’s Remarks, 11/21/2025, pages 15-16]
In response to the Applicant’s argument with respect to the limitation “presenting the one or more candidate subtasks generated based on the task and each supplemental task of the set to the user through an interface generated by the task management system,” the Examiner notes the limitation being argued by Applicant as being newly amended to the claims in the response filed 11/21/2025, which has been addressed in the updated rejection below. Applicant’s argument has been considered, but it pertains to amendments to independent claim 1 that are believed to be addressed via the new ground of rejection under §103 set forth in the instant Office action, which incorporates new citations and new references to address the amended limitations in claim 1 and supports a conclusion of obviousness of the amended claims. Accordingly, the amendment and supporting arguments are believed to be fully addressed via the updated ground of rejection set forth under §103 below.
Applicant submits “Sodhi makes no disclosure or suggestion of "receiving a selection of a candidate subtask from the user via the interface" and "storing the selected candidate subtask in association with the task at the task management system," as claimed.” [Applicant’s Remarks, 11/21/2025, page 16]
In response to the Applicant’s argument with respect to the limitations “receiving a selection of a candidate subtask from the user via the interface; and storing the selected candidate subtask in association with the task at the task management system,” the Examiner notes the limitations being argued by Applicant as being newly amended to the claims in the response filed 11/21/2025, which have been addressed in the updated rejection below. Applicant’s argument has been considered, but it pertains to amendments to independent claim 1 that are believed to be addressed via the new ground of rejection under §103 set forth in the instant Office action, which incorporates new citations and new references to address the amended limitations in claim 1 and supports a conclusion of obviousness of the amended claims. Accordingly, the amendment and supporting arguments are believed to be fully addressed via the updated ground of rejection set forth under §103 below.
Applicant submits “Herman does not present "the one or more candidate subtasks generated based on the task and each supplemental task of the set to the user through an interface generated by the task management system," as recited by amended independent claim 1.” [Applicant’s Remarks, 11/21/2025, page 17]
In response to the Applicant’s argument with respect to the limitations “presenting the one or more candidate subtasks generated based on the task and each supplemental task of the set to the user through an interface generated by the task management system,” the Examiner notes the limitation being argued by Applicant as being newly amended to the claims in the response filed 11/21/2025, which has been addressed in the updated rejection below. Applicant’s argument has been considered, but it pertains to amendments to independent claim 1 that are believed to be addressed via the new ground of rejection under §103 set forth in the instant Office action, which incorporates new citations and new references to address the amended limitations in claim 1 and supports a conclusion of obviousness of the amended claims. Accordingly, the amendment and supporting arguments are believed to be fully addressed via the updated ground of rejection set forth under §103 below.
Applicant submits “Herman also makes no disclosure or suggestion of "receiving a selection of a candidate subtask from the user via the interface" and "storing the selected candidate subtask in association with the task at the task management system," as claimed.” [Applicant’s Remarks, 11/21/2025, pages 17-18]
In response to the Applicant’s argument with respect to the limitations “receiving a selection of a candidate subtask from the user via the interface; and storing the selected candidate subtask in association with the task at the task management system,” the Examiner notes the limitations being argued by Applicant as being newly amended to the claims in the response filed 11/21/2025, which have been addressed in the updated rejection below. Applicant’s argument has been considered, but it pertains to amendments to independent claim 1 that are believed to be addressed via the new ground of rejection under §103 set forth in the instant Office action, which incorporates new citations and new references to address the amended limitations in claim 1 and supports a conclusion of obviousness of the amended claims. Accordingly, the amendment and supporting arguments are believed to be fully addressed via the updated ground of rejection set forth under §103 below.
16. Applicant’s remaining arguments either logically depend from the above-rejected arguments, in which case they too are unpersuasive for the reasons set forth above, or they are
directed to features which have been newly added via amendment. Therefore, this is now the Examiner's first opportunity to consider these limitations and as such any arguments regarding these limitations would be inappropriate since they have not yet been examined. A full rejection of these limitations will be presented later in this Office Action.
Claim Rejections - 35 USC § 101
17. 35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
18. Claims 1-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to non-patentable subject matter. The claims are directed to an abstract idea without significantly more.
19. Claims 1-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more. The eligibility analysis in support of these findings is provided below, in accordance with MPEP 2106.
With respect to Step 1 of the eligibility inquiry (as explained in MPEP 2106), it is first noted that the claimed method (claims 1-10), computer program product (claims 11-19), and method (claim 20) are directed to potentially eligible categories of subject matter (i.e., process, article of manufacture, and process, respectively), and therefore claims 1-20 satisfy Step 1 of the eligibility inquiry.
With respect to Step 2A Prong One, it is next noted that the claims recite an abstract idea that falls into the “Certain Methods of Organizing Human Activity” abstract idea set forth in MPEP 2106 because the claims recite steps for managing tasks, which encompasses activity for managing personal behavior or relationships or interactions. With respect to independent claims 1/11/20, the limitations reciting the abstract idea are indicated in bold below: receiving, at the task management system, a task from a user, the task specifying an action for performance by a performing user; determining, by the task management system, a list including the task; selecting, at the task management system, a set of supplemental tasks based on the list including the task, based on creation dates of tasks associated with the list, and based on measures of similarity between an embedding determined for the task and embeddings determined for tasks associated with the list; generating, by the task management system, a prompt for a generative model including the task and each supplemental task of the set, the prompt including at least a title of the task and a title of each supplemental task of the set; generating one or more candidate subtasks for the task by applying the generative model to the prompt; presenting the one or more candidate subtasks generated based on the task and each supplemental task of the set to the user through an interface generated by the task management system; receiving a selection of a candidate subtask from the user via the interface; and storing the selected candidate subtask in association with the task at the task management system. These steps cover organizing human activity because the received information directly pertains to user task scheduling. Independent claims 11 and 20 recite similar limitations as set forth in claim 1 and are therefore found to recite the same abstract idea as claim 1.
Therefore, because the limitations above set forth activities falling within the “Certain methods of organizing human activity” abstract idea grouping described in MPEP 2106, the additional elements recited in the claims are further evaluated, individually and in combination, under Step 2A Prong Two and Step 2B below.
With respect to Step 2A Prong Two, the judicial exception is not integrated into a practical application. The additional elements are: a task management system, processor, a non-transitory computer readable medium, a generative model, an interface generated by the task management system (claim 1), a non-transitory computer readable storage medium having instructions encoded thereon, a processor, a task management system including the processor, a generative model, and an interface generated by the task management (claim 11), a task management system, a processor, a non-transitory computer readable medium, a generative model, an interface generated by the task management system (claim 20). These additional elements have been evaluated, but fail to integrate the abstract idea into a practical application because they amount to using generic computing elements or computer-executable instructions (software) to perform the abstract idea, similar to adding the words “apply it” (or an equivalent), which merely serves to link the use of the judicial exception to a particular technological environment. See MPEP 2106.05(f) and 2106.05(h). Furthermore, these additional elements fail to integrate the abstract idea into a practical application because they fail to provide an improvement to the functioning of a computer or to any other technology or technical field, fail to apply the exception with a particular machine, fail to apply the judicial exception to effect a particular treatment or prophylaxis for a disease or medical condition, fail to effect a transformation of a particular article to a different state or thing, and fail to apply/use the abstract idea in a meaningful way beyond generally linking the use of the judicial exception to a particular technological environment.”). Even if the step for receiving is evaluated as an additional element, this activity encompasses, at most, insignificant extra-solution activity, which is not indicative of a practical application, as noted in MPEP 2106.05(g), and is not enough to add significantly more since it is well-understood and conventional activity, as noted in MPEP 2106.05(d)
Accordingly, because the Step 2A Prong One and Prong Two analysis resulted in the conclusion that the claims are directed to an abstract idea, additional analysis under Step 2B of the eligibility inquiry must be conducted in order to determine whether any claim element or combination of elements amount to significantly more than the judicial exception.
With respect to Step 2B of the eligibility inquiry, it has been determined that the claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception. The additional elements are: a task management system, processor, a non-transitory computer readable medium, a generative model, an interface generated by the task management system (claim 1), a non-transitory computer readable storage medium having instructions encoded thereon, a processor, a task management system including the processor, a generative model, and an interface generated by the task management (claim 11), a task management system, a processor, a non-transitory computer readable medium, a generative model, an interface generated by the task management system (claim 20). The additional elements have been fully considered, but fail to add significantly more because they merely serve to tie the invention to a particular operating environment (i.e., computer-based implementation) by describing the use of generic computing elements to implement the claimed invention, though at a very high level of generality and without imposing meaningful limitation on the scope of the claim, similar to simply saying "apply it” or “apply it using a general purpose computer,” which is not enough to transform an abstract idea into eligible subject matter. Notably, Applicant’s Specification describes generic off-the-shelf computing elements for implementing the claimed invention and suggests that virtually any generic computing devices could be used to implement the invention (See, e.g., Specification paragraphs [0016]”). Therefore, these additional elements describe generic computing elements that merely serve to tie the abstract idea to a particular operating environment, which does not add significantly more to the abstract idea. See, e.g., Alice Corp., 134 S. Ct. 2347, 110 USPQ2d 1976. Even if the step for receiving is evaluated as an additional element, this activity encompasses, at most, insignificant extra-solution activity, which is not enough to add significantly more since it is well-understood and conventional activity, as noted in MPEP 2106.05(d).
Similarly, with respect to the “storing” step, even if considered as an additional element, when evaluated under Step 2A Prong Two and Step 2B, amounts to insignificant extra-solution activity, which does not amount to a practical application (MPEP 2106.05(g)), nor add significantly more because such activity has been recognized as well-understood, routine, and conventional and thus insufficient to add significantly more to the abstract idea. See MPEP 2106.05(d). “The courts have recognized the following computer functions as well‐understood, routine, and conventional functions when they are claimed in a merely generic manner (e.g., at a high level of generality) or as insignificant extra-solution activity: iv. Storing and retrieving information in memory, Versata Dev. Group, Inc. v. SAP Am., Inc., 793 F.3d 1306, 1334, 115 USPQ2d 1681, 1701 (Fed. Cir. 2015); OIP Techs., 788 F.3d at 1363, 115 USPQ2d at 1092-93; v. Electronically scanning or extracting data from a physical document, Content Extraction and Transmission, LLC v. Wells Fargo Bank, 776 F.3d 1343, 1348, 113 USPQ2d 1354, 1358 (Fed. Cir. 2014) (optical character recognition).” See MPEP 2106.05(d).
In addition, when taken as an ordered combination, the ordered combination adds nothing that is not already present as when the elements are taken individually. There is no indication that the combination of elements integrate the abstract idea into a practical application. Their collective functions merely provide generic computer implementation. Therefore, when viewed as a whole, these additional claim elements do not provide meaningful limitations to transform the abstract idea into a practical application of the abstract idea or that the ordered combination amounts to significantly more than the abstract idea itself.
Dependent claims 2-10 recite the same abstract idea as recited in the independent claims, and when evaluated under Step 2A Prong One of the eligibility inquiry, merely recite further details of the same abstract idea recited in the independent claims accompanied by, at most, the involvement of the same generic computing elements as the independent claims which, as noted above, are not sufficient to amount to a practical application or significantly more than the abstract idea itself. In particular, dependent claims 2-10 recite “wherein selecting the set of supplemental tasks based on the list including the task, based on creation dates of tasks associated with the list, and based on measures of similarity between an embedding determined for the task and embeddings determined for tasks associated with the list comprises: selecting tasks associated with the list including the task and created by the user as the set of supplemental tasks,” “wherein selecting the set of supplemental tasks based on the list including the task, based on creation dates of tasks associated with the list, and based on measures of similarity between an embedding determined for the task and embeddings determined for tasks associated with the list comprises: selecting tasks associated with the list including the task, created by the user, and created within a threshold amount of time from a time when the task management system received the task from the user as the set of supplemental tasks,” “wherein selecting the set of supplemental tasks based on the list including the task, based on creation dates of tasks associated with the list, and based on measures of similarity between an embedding determined for the task and embeddings determined for tasks associated with the list comprises: selecting additional tasks associated with the list including the task and created by the user; generating a ranking of the additional tasks based on times when the additional tasks were created relative to a time when the task management system received the task from the user, where additional tasks more recently created relative to the time when the task management system received the task have higher positions in the ranking; and selecting additional tasks having at least a threshold position in the ranking as the set of supplemental tasks,” “wherein selecting, the set of supplemental tasks based on the list including the task, based on creation dates of tasks associated with the list, and based on measures of similarity between an embedding determined for the task and embeddings determined for tasks associated with the list comprises: generating the embedding for the task based on received information from the user describing the task; determining embeddings for additional tasks maintained; and selecting additional tasks having embeddings with at least a threshold measure of similarity to the embedding for the task as the set of supplemental tasks,” “wherein each additional task was created by the user,” “wherein each additional task is associated with the list including the task,” “wherein each additional task was created within a threshold amount of time of a time when the task management system received the task,” “wherein the embedding for the task is determined based on a name of the task and a description of the task,” “further comprising: modifying the selected candidate subtask in response to the task management system receiving an input from the user,” however these limitations cover organizing human activity since they flow directly from task management, which encompasses activity for managing personal behavior or relationships or interactions. The other dependent claims have been fully considered as well; however these claims are also directed to the abstract idea itself without integrating it into a practical application and implemented by, at most, a general purpose computer that serves to tie the idea to a particular technological environment, which does not add significantly more to the claims. The additional elements recited in the dependent claims include the task management system. However, this element is recited at a high level of generality and fails to yield any discernible improvement to the computer or to any technology, nor set forth any additional function or result that provided meaningful limitation beyond linking the abstract idea to a particular technological environment (i.e., automated/computing environment), and thus fail to integrate the abstract idea into a practical application. When evaluated under Step 2A Prong Two and Step 2B, the additional elements do not amount to a practical application or significantly more since they merely require generic computing devices (or computer-implemented instructions/code) which as noted in the discussion of the independent claims above is not enough to render the claims as eligible.
The ordered combination of elements in the dependent claims (including the limitations inherited from the parent claim(s)) add nothing that is not already present as when the elements are taken individually. There is no indication that the combination of elements improves the functioning of a computer or improves any other technology. Their collective functions merely provide generic computer implementation. Accordingly, the subject matter encompassed by the dependent claims fails to amount to a practical application or significantly more than the abstract idea itself.
For more information, see MPEP 2106.
Claim Rejections - 35 USC § 103
20. In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
21. The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102 of this title, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
22. The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
23. Claims 1-3, 5-13, and 15-20 are rejected under 35 U.S.C. 103 as being unpatentable over Sodhi et al., Pub. No.: US 2024/0386216 A1, [hereinafter Sodhi], in view of Herman et al., Pub. No.: US 2025/0278175 A1, [hereinafter Herman], in view of Gruber et al., Pub. No.: US 2012/0311583 A1, [hereinafter Gruber], in further view of Fowler et al., Pub. No.: US 2018/0129371 A1, [hereinafter Fowler].
As per claim 1, Sodhi teaches a method, performed at a task management system comprising a processor and a non-transitory computer readable medium (paragraphs 0006, 0014), comprising:
receiving, at the task management system, a task from a user, the task specifying an action for performance by a performing user (paragraph 0006, discussing a computer-implemented method, including: receiving instruction text including instructions for accomplishing a task; paragraph 0038, discussing that a user may submit a text question to a Large Language Model (LLM) to receive a response or may provide text instructions to an LLM to perform a task);
determining, by the task management system, a list (paragraph 0095, discussing that a task prompt template may be constructed to focus on higher level parts of a task. In some implementations, a task prompt template may instruct a language model to select a subtask as a next action to be performed. In some implementations, a task prompt template may instruct a language model to provide a sequence of subtasks to be performed to complete the task. In some implementations, a task prompt template may only allow a language model to select from a list of possible subtasks. In some implementations, a task prompt template may allow a language model to select from a list of possible subtasks or from a list of possible web page operations to be performed on a web page; paragraph 0103, discussing a list of subtasks to be performed to complete the task);
generating, by the task management system, a prompt for a generative model including the task (paragraph 0014, discussing selecting a task prompt template using the instruction text…; create a first language model prompt using (a) the task prompt template; paragraph 0015, discussing wherein the task prompt template includes a first task prompt; paragraph 0063, discussing a prompt template that may be used to generate a prompt that may be submitted to a language model; paragraph 0065, discussing that the prompt template may have general instructions that provide high-level information about the task to be completed);
generating one or more candidate subtasks for the task by applying the generative model to the prompt (paragraph 0046, discussing that to improve the ability of a language model to perform complex tasks, the complex task may be broken down into a main task and subtasks. The main task may use a task prompt template to provide higher level guidance for how to complete the task and the subtask may use a subtask prompt template to indicate how to implement portions of the task; paragraph 0087, discussing that it may be desired to modify the above techniques to allow a task to be broken down into subtasks. A language model may be used to select a sequence of subtasks to be performed…More than two levels are possible, such as breaking up a subtask into subsubtasks, and so forth. Breaking a task into subtasks may be desired for a variety of reasons; paragraph 0095, discussing that a task prompt template may instruct a language model to select a subtask as a next action to be performed. In some implementations, a task prompt template may instruct a language model to provide a sequence of subtasks to be performed to complete the task. In some implementations, a task prompt template may only allow a language model to select from a list of possible subtasks. In some implementations, a task prompt template may allow a language model to select from a list of possible subtasks);
presenting the one or more candidate subtasks to the user generated based on the task (paragraph 0061, discussing that the response of the language model may include any appropriate text, such as one or more of the following: text to be presented to a user;…; or an indication to perform other actions with the language model, such as using a sub-task prompt template; paragraph 0085, discussing that a description of the completed task may be presented to the user; paragraph 0095, discussing that a task prompt template may instruct a language model to select a subtask as a next action to be performed. In some implementations, a task prompt template may instruct a language model to provide a sequence of subtasks to be performed to complete the task. In some implementations, a task prompt template may only allow a language model to select from a list of possible subtasks. In some implementations, a task prompt template may allow a language model to select from a list of possible subtasks; paragraph 0099, discussing a prompt template generation component that may store the generated task prompt templates in a task prompt templates data store and may store the generated subtask prompt templates in a subtask prompt templates data store; paragraph 0103, discussing providing a list of subtasks to be performed to complete the task; paragraph 0107); and
receiving a selection of a candidate subtask (paragraph 0095, discussing that a task prompt template may instruct a language model to select a subtask as a next action to be performed. In some implementations, a task prompt template may instruct a language model to provide a sequence of subtasks to be performed to complete the task. In some implementations, a task prompt template may only allow a language model to select from a list of possible subtasks. In some implementations, a task prompt template may allow a language model to select from a list of possible subtasks or from a list of possible web page operations to be performed on a web page; paragraph 0103, discussing a list of subtasks to be performed to complete the task).
While Sodhi describes receiving a selection of a candidate subtask (paragraphs 0095, 0103) and storing task prompt templates (paragraph 0099, discussing that the prompt template generation component may store the generated task prompt templates in task prompt templates data store and may store the generated subtask prompt templates in subtask prompt templates data store), Sodhi does not explicitly teach determining, by the task management system, a list including the task; selecting, at the task management system, a set of supplemental tasks based on the list including the task, based on creation dates of tasks associated with the list, and based on measures of similarity between an embedding determined for the task and embeddings determined for tasks associated with the list; generating, by the task management system, a prompt for a generative model including the task and each supplemental task of the set, the prompt including at least a title of the task and a title of each supplemental task of the set; presenting the one or more candidate subtasks generated based on the task and each supplemental task of the set to the user through an interface generated by the task management system; receiving a selection of a candidate subtask from the user via the interface; and storing the selected candidate subtask in association with the task at the task management system. Herman in the analogous art of large language models teaches:
determining, by the task management system, a list including the task (paragraph 0004, discussing determining, based on the plurality of tasks, an ordered task list associated with a set of task dependency criteria, wherein each task of the ordered task list includes a respective task dependency criteria; displaying, via the display generation component, a representation of the ordered task list; paragraph 0008, discussing developing and executing an ordered task list using LLM (Large Language Model) functionality provides an enhanced interface for user-device interaction);
selecting, at the task management system, a set of supplemental tasks based on the list including the task (paragraph 0311, discussing that once the task(s) are determined, each task is assessed as to whether the task satisfies management criteria…In some examples, LLM model tuning is used to determine capabilities of computer system. Specifically, a task-specific dataset may be applied to LLM 910, such that the LLM architecture and LLM parameters are optimized in order to handle requests directed to a specific domain or area of interest. The fine-tuned LLM may then be used to assess system capabilities. Tasks satisfying the management criteria are then be added to an ordered task list. Any tasks not satisfying the management criteria may be subject to additional follow-up. For instance, the user may be prompted to modify the task request and/or provide additional information in order to facilitate proper processing of the task (e.g., the user may be prompted “Did you mean BestNewsPage.com?”); paragraph 0312, discussing that each task of the ordered task list may include various task dependency criteria. In general, the task dependency criteria may indicate which (if any) tasks must be performed prior to a respective task being performed. For example, in order to perform the fourth task “(4) Compose message to John's contact identifier including link to identified headline story,” the second task “(2) Identify link to headline story on ‘BestNews.com’ within web browser application” must first be performed. Likewise, in order to perform the second task “(2) Identify link to headline story on ‘BestNews.com’ within web browser application” the first task “(1) Navigate to ‘BestNews.com’ using web browser application” must first be performed. Generally, tasks that are dependent on the performance of other tasks within the ordered task list may be referred to as synchronous tasks, whereas tasks that are not dependent on the performance of any other tasks within the ordered task list may be referred to as asynchronous tasks; paragraph 0417, discussing determining (e.g., using an AI process or a generative AI process), based on the plurality of tasks, an ordered task list associated with a set of task dependency criteria, wherein each task of the ordered task list includes a respective task dependency criteria (e.g., each task includes information on whether the task is dependent on the result of other tasks or is independent of the result of other tasks). In some examples, based on the parsed commands, a set of tasks for satisfying the commands is determined (e.g., using an AI process or a generative AI process). In some examples, a determination (e.g., using an AI process or a generative AI process) is made whether each task of the set of tasks is manageable; paragraph 0429), and based on measures of similarity between an embedding determined for the task and embeddings determined for tasks associated with the list (paragraph 0204, discussing that a comparison between the characteristic intensity and one or more thresholds is used to determine whether or not to perform one or more operations; paragraph 0246, discussing that the natural language processing module receives the candidate text representations…, and for each candidate representation, determines what nodes are implicated by the words in the candidate text representation. In some examples, if a word or phrase in the candidate text representation is found to be associated with one or more nodes in ontology 760, the word or phrase “triggers” or “activates” those nodes. Based on the quantity and/or relative importance of the activated nodes, the natural language processing module selects one of the actionable intents as the task that the user intended the digital assistant to perform. In some examples, the domain that has the most “triggered” nodes is selected. In some examples, the domain having the highest confidence value (e.g., based on the relative importance of its various triggered nodes) is selected. In some examples, the domain is selected based on a combination of the number and the importance of the triggered nodes. In some examples, additional factors are considered in selecting the node as well, such as whether the digital assistant has previously correctly interpreted a similar request from a user; paragraph 0307, discussing that the classification type included in the LLM output indicates whether a user request (e.g., represented by the input embedding) corresponds to a request to initiate a plan. For example, the LLM is trained to classify an input as a request to initiate a plan (a planning request) or as not a request to initiate a plan (a non-planning request). Generally, a plan includes, at least initially, one or more tasks suggested by the computer system for completion of the plan. Non-limiting examples of planning requests include a request to plan a vacation (e.g., “help me plan a trip to Tokyo”), a request for an exercise plan (e.g., “help me run a 6 minute mile”), a request for a diet plan (e.g., “help me eat healthier”), a request to plan an event (e.g., “help me plan Thanksgiving dinner”),…, and other requests to achieve a user-specified goal. The computer system interprets such planning requests as requests to generate respective suggestion(s) associated with the plan, e.g., suggestions for steps/tasks associated with planning a trip to Tokyo or suggestions for steps/tasks associated with running a mile in 6 minutes or less...Whether a particular request is classified as a planning request or as a non-planning request can depend on the particular implementation of the LLM. As described below with respect to FIGS. 12A-12Y, in some examples, computer system performs different actions responsive to a request depending on whether it is classified as a planning request or as a non-planning request. Accordingly, in some examples, planning requests and non-planning requests are distinguished by the differing actions computer system performs in response to receiving the respective request; paragraph 0308, discussing that the LLM output is then passed to task flow module which performs various functions to satisfy the commands. The task flow module may also determine that additional information is needed in order to satisfy the commands, such that the task flow module passes one or more output representations back to the input embedding layer. In some examples, the output from the task flow module may first be passed to the tokenization layer, or alternatively, may be passed directly to the LLM (e.g., depending on the format of the output from task flow module). The task flow module may cause one of more outputs to be provided at the end of processing or throughout processing; paragraph 0306).
generating, by the task management system, a prompt for a generative model including the task and each supplemental task of the set, the prompt including at least a title of the task and a title of each supplemental task of the set (paragraph 0314, discussing that as tasks within the ordered task list are performed, context information may be accumulated and used for optimized task performance. For example, the user may request a local dinner reservation for four specific people along with a request to send the reservation confirmation to the attendees. While performing the ordered task list, a determination is made that there are two locations for the respective restaurant that are generally reachable by the user (e.g., within 10 miles). Context information regarding each restaurant may then be gathered in order to determine whether additional follow-up with the user is needed. For example, the additional context information may include restaurant ratings, reservation availability, the user's preference regarding either location, and the like. To the extent a determination cannot be made as to which restaurant the user intends to dine at, an additional task can be created (e.g., “Provide prompt to user asking whether the user prefers to dine at Restaurant X or Restaurant Y). The additional task can then be inserted into the ordered task list at the appropriate slot based on any relevant task dependency criteria; paragraph 0385, discussing that the planning user interface includes one or more graphical depictions of one or more planning tasks with which the DA currently assists the user. Planning user interface 1224 includes planning object 1226 that device 1200 added based on processing speech input 1212. Planning user interface 1224 further includes planning object 1228. Device 1200 added planning object 1228 (that corresponds to a task of planning a trip to Tokyo) based on processing speech input 1238, as detailed below with respect to FIGS. 12L-12Q. In some examples, planning objects 1226 and 1228 each include text describing their respective plans (e.g., the title “Tokyo trip” and the title “eat healthy and lose weight”) and optionally, one or more graphics for their respective plans (e.g., an image of Tokyo and image(s) associated with exercise and healthy eating); paragraph 0417, discussing determining (e.g., using an AI process or a generative AI process), based on the plurality of tasks, an ordered task list associated with a set of task dependency criteria, wherein each task of the ordered task list includes a respective task dependency criteria (e.g., each task includes information on whether the task is dependent on the result of other tasks or is independent of the result of other tasks). In some examples, based on the parsed commands, a set of tasks for satisfying the commands is determined (e.g., using an AI process or a generative AI process). In some examples, a determination (e.g., using an AI process or a generative AI process) is made whether each task of the set of tasks is manageable; paragraph 0429, discussing that in accordance with a determination (e.g., using an AI process or a generative AI process) that additional information is required in order to perform a respective task of the ordered last list, identifying (e.g., using an AI process or a generative AI process) an additional task based on the additional information and adding the additional task to the ordered task list (e.g., task prompting user for additional information); paragraphs 0389, 0484);
presenting the one or more candidate subtasks generated based on the task and each supplemental task of the set to the user through an interface generated by the task management system (paragraph 0004, discussing determining, based on the plurality of tasks, an ordered task list associated with a set of task dependency criteria, wherein each task of the ordered task list includes a respective task dependency criteria; displaying, via the display generation component, a representation of the ordered task list; performing each task of the ordered task list based on the set of task dependency criteria; while performing each task: updating the display, via the display generation component, of the representation of the ordered task list to indicate completion status of each task of the ordered task list; and displaying, via the display generation component, at least one graphical object corresponding to a respective task of the ordered task list; paragraph 0396, discussing that the response object includes suggested tasks corresponding to the plan for the Tokyo trip. In some examples, device 1200 orders suggested tasks 1252 based on dependencies between the suggested tasks. A dependency between two suggested tasks indicates that a result associated with performance of one of the tasks is required to initiate performance of the other task. In some examples, the DA (e.g., LLM unit 812) determines the dependencies between the suggested tasks of the plan when generating the plan…Displaying suggested tasks of a plan in an order that is based on their respective dependencies may assist the user in determining a proper order in which to provide information to device 1200 to complete the plan, which in turn allows the user to use device 1200 more quickly and efficiently; paragraph 0465, discussing that process 1400 further includes: while displaying the second set of suggestions in the region, detecting, via the one or more input devices, a user input (corresponding to a selection of the second set of suggestions (e.g., 1122); and in response to detecting the user input corresponding to the selection of the second set of suggestions: in accordance with a determination (e.g., using an AI process or a generative AI process) that the user input corresponding to the selection of the second set of suggestions corresponds to movement in a first direction, displaying, via the display generation component, a fourth set of suggestions different from the second set of suggestions, wherein the fourth set of suggestions is determined based on the first content; paragraph 0484, discussing detecting, via the one or more input devices, a user input corresponding to a selection of the ongoing plan graphical object; and in response to detecting the user input corresponding to a selection of the ongoing plan graphical object, displaying, via the display generation component, a user interface corresponding to the plan. In some examples, the user interface corresponding to the plan includes more detail about the plan than the ongoing plan graphical object. For example, the user interface includes the steps and/or suggestions of the plan while the ongoing plan graphical object includes a title of the plan and a graphical element corresponding to the plan; paragraphs 0371, 0403); and
receiving a selection of a candidate subtask from the user via the interface (paragraph 0465, discussing that process 1400 further includes: while displaying the second set of suggestions in the region, detecting, via the one or more input devices, a user input (corresponding to a selection of the second set of suggestions (e.g., 1122); and in response to detecting the user input corresponding to the selection of the second set of suggestions: in accordance with a determination (e.g., using an AI process or a generative AI process) that the user input corresponding to the selection of the second set of suggestions corresponds to movement in a first direction, displaying, via the display generation component, a fourth set of suggestions different from the second set of suggestions, wherein the fourth set of suggestions is determined based on the first content; paragraph 0484, discussing detecting, via the one or more input devices, a user input corresponding to a selection of the ongoing plan graphical object; and in response to detecting the user input corresponding to a selection of the ongoing plan graphical object, displaying, via the display generation component, a user interface corresponding to the plan. In some examples, the user interface corresponding to the plan includes more detail about the plan than the ongoing plan graphical object. For example, the user interface includes the steps and/or suggestions of the plan while the ongoing plan graphical object includes a title of the plan and a graphical element corresponding to the plan; paragraph 0468).
Sodhi is directed towards automation of tasks using language model prompts. Herman is directed to systems and techniques for incorporating large language models into intelligent automated assistants. Therefore, they are deemed to be analogous as they both are directed towards solutions for task management and large language models. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine Sodhi with Herman because the references are analogous art because they are both directed to solutions for task management and generative models, which falls within applicant’s field of endeavor (task management system), and because modifying Sodhi to include Herman’s features for determining, by the task management system, a list including the task; selecting, at the task management system, a set of supplemental tasks based on the list including the task, and based on measures of similarity between an embedding determined for the task and embeddings determined for tasks associated with the list; generating, by the task management system, a prompt for a generative model including the task and each supplemental task of the set, the prompt including at least a title of the task and a title of each supplemental task of the set; presenting the one or more candidate subtasks generated based on the task and each supplemental task of the set to the user through an interface generated by the task management system; and receiving a selection of a candidate subtask from the user via the interface, in the manner claimed, would serve the motivation of allowing more efficient and accurate management of user requested plans (Herman at paragraph 0018); and further obvious because the claimed invention is merely a combination of old elements, and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable.
The Sodhi-Herman combination does not explicitly teach selecting, at the task management system, a set of supplemental tasks based on creation dates of tasks associated with the list; the prompt including at least a title of the task and a title of each supplemental task of the set; and storing the selected candidate subtask in association with the task at the task management system. Gruber in the analogous art of task generation systems teaches:
selecting, at the task management system, a set of supplemental tasks based on creation dates of tasks associated with the list (paragraph 0032, discussing that a suggestion engine is operable to determine whether a candidate task received from a tasks source should be suggested to the user as a task to perform at a given time and/or location. From all of the candidate task items that may be presented to the user at any given time, the suggestion engine filters those task items to a manageable subset based on the user's existing task items, prior acceptances/rejections of suggested task items, and the prior actions of the user. For example, if a user's calendar includes an event for an upcoming birthday, a suggested task is created that the person whose birthday is coming up should be called prior to that date; paragraph 0035, discussing that a preview generator is operable to generate previews for entities associated with a suggested task (or a selected task); paragraph 0048, discussing that upon selection of the task creation control, the “volunteer registration” task is created as a task item for “tonight”. In another aspect, the user selects the task creation option and identifies the objects, persons, and times relevant to that task; paragraph 0049, discussing that upon selection of the Task creation option, the task item is created in relation to the content item from which it was created and is displayed in a calendar. For example, the system identifies the “volunteer registration” task in the email and that the task needs to be completed tonight—relative to the day of creation or a date in the content item. The system creates the task and displays the task in the calendar application for completion tonight. In one example, the user is provided with an option whether to accept the task or to make any desired changes to the task such as for example, change the date and or time, reassigned the task, etc.; paragraph 0052).
The Sodhi-Herman combination describes features related to task management. Gruber is directed to a system and method for generating and processing task items. Therefore, they are deemed to be analogous as they both are directed towards solutions for task management and large language models. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the Sodhi-Herman combination with Gruber because the references are analogous art because they are both directed to solutions for task management and generative models, which falls within applicant’s field of endeavor (task management system), and because modifying the Sodhi-Herman combination to include Gruber’s feature for selecting, at the task management system, a set of supplemental tasks based on creation dates of tasks associated with the list, in the manner claimed, would serve the motivation of assisting a user in managing his/her tasks (Gruber at paragraph 0028); and further obvious because the claimed invention is merely a combination of old elements, and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable.
The Sodhi-Herman-Gruber combination does not explicitly teach storing the selected candidate subtask in association with the task at the task management system. However, Fowler in the analogous art of task management systems teaches this concept. Fowler teaches:
storing the selected candidate subtask in association with the task at the task management system (paragraph 0038, discussing that a relational store stores the relations observed for the creation of task items so that dynamic context can be provided to the user when the task is suggested to the user at a later date. For example, when the user manually or a system automatically creates a task item, the task is parsed to locate entities (e.g., persons involved, objects to be acted on) and recent actions (e.g., actions taken in the last m minutes) that may relate to the task item. For example, if the user receives a message containing the phrase “profit sharing plan” and creates a task that also include that phrase, a relationship between the task and the message will be formed and stored in the relational store; paragraph 0060, discussing that the relationships for the task item are stored in a relationship store).
The Sodhi-Herman-Gruber combination describes features related to task management. Fowler is directed to a system and method for presenting and manipulating task items. Therefore, they are deemed to be analogous as they both are directed towards solutions for task management. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the Sodhi-Herman-Gruber combination with Fowler because the references are analogous art because they are both directed to solutions for task management and generative models, which falls within applicant’s field of endeavor (task management system), and because modifying the Sodhi-Herman-Gruber combination to include Fowler’s feature for storing the selected candidate subtask in association with the task at the task management system, in the manner claimed, would serve the motivation of enhancing the determinations of which candidate tasks discovered from task sources are to be presented (Fowler at paragraph 0031); and further obvious because the claimed invention is merely a combination of old elements, and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable.
As per claim 2, the Sodhi-Herman-Gruber-Fowler combination teaches the method of claim 1. Although not explicitly taught by Sodhi, Herman in the analogous art of task management systems teaches wherein selecting, at the task management system, the set of supplemental tasks based on the list including the task and based on measures of similarity between an embedding determined for the task and embeddings determined for tasks associated with the list comprises: selecting tasks associated with the list including the task and created by the user as the set of supplemental tasks (paragraph 0311, discussing that once the task(s) are determined, each task is assessed as to whether the task satisfies management criteria…In some examples, LLM model tuning is used to determine capabilities of computer system. Specifically, a task-specific dataset may be applied to LLM 910…The fine-tuned LLM may then be used to assess system capabilities. Tasks satisfying the management criteria are then be added to an ordered task list. Any tasks not satisfying the management criteria may be subject to additional follow-up. For instance, the user may be prompted to modify the task request and/or provide additional information in order to facilitate proper processing of the task (e.g., the user may be prompted “Did you mean BestNewsPage.com?”); paragraph 0312, discussing that each task of the ordered task list may include various task dependency criteria. In general, the task dependency criteria may indicate which (if any) tasks must be performed prior to a respective task being performed. For example, in order to perform the fourth task “(4) Compose message to John's contact identifier including link to identified headline story,” the second task “(2) Identify link to headline story on ‘BestNews.com’ within web browser application” must first be performed. Likewise, in order to perform the second task “(2) Identify link to headline story on ‘BestNews.com’ within web browser application” the first task “(1) Navigate to ‘BestNews.com’ using web browser application” must first be performed. Generally, tasks that are dependent on the performance of other tasks within the ordered task list may be referred to as synchronous tasks, whereas tasks that are not dependent on the performance of any other tasks within the ordered task list may be referred to as asynchronous tasks; paragraph 0417, discussing determining (e.g., using an AI process or a generative AI process), based on the plurality of tasks, an ordered task list associated with a set of task dependency criteria, wherein each task of the ordered task list includes a respective task dependency criteria (e.g., each task includes information on whether the task is dependent on the result of other tasks or is independent of the result of other tasks). In some examples, based on the parsed commands, a set of tasks for satisfying the commands is determined (e.g., using an AI process or a generative AI process). In some examples, a determination (e.g., using an AI process or a generative AI process) is made whether each task of the set of tasks is manageable).
Sodhi is directed towards automation of tasks using language model prompts. Herman is directed to systems and techniques for incorporating large language models into intelligent automated assistants. Therefore, they are deemed to be analogous as they both are directed towards solutions for task management and large language models. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine Sodhi with Herman because the references are analogous art because they are both directed to solutions for task management and generative models, which falls within applicant’s field of endeavor (task management system), and because modifying Sodhi to include Herman’s feature for including wherein selecting, at the task management system, the set of supplemental tasks based on the list including the task and based on measures of similarity between an embedding determined for the task and embeddings determined for tasks associated with the list comprises: selecting tasks associated with the list including the task and created by the user as the set of supplemental tasks, in the manner claimed, would serve the motivation of allowing more efficient and accurate management of user requested plans (Herman at paragraph 0018); and further obvious because the claimed invention is merely a combination of old elements, and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable.
The Sodhi-Herman combination does not explicitly teach selecting, at the task management system, the set of supplemental tasks based on the list including the task, based on creation dates of tasks associated with the list. Gruber in the analogous art of task generation systems teaches:
selecting, at the task management system, the set of supplemental tasks based on the list including the task, based on creation dates of tasks associated with the list (paragraph 0032, discussing that a suggestion engine is operable to determine whether a candidate task received from a tasks source should be suggested to the user as a task to perform at a given time and/or location. From all of the candidate task items that may be presented to the user at any given time, the suggestion engine filters those task items to a manageable subset based on the user's existing task items, prior acceptances/rejections of suggested task items, and the prior actions of the user. For example, if a user's calendar includes an event for an upcoming birthday, a suggested task is created that the person whose birthday is coming up should be called prior to that date; paragraph 0035, discussing that a preview generator is operable to generate previews for entities associated with a suggested task (or a selected task); paragraph 0048, discussing that upon selection of the task creation control, the “volunteer registration” task is created as a task item for “tonight”. In another aspect, the user selects the task creation option and identifies the objects, persons, and times relevant to that task; paragraph 0049, discussing that upon selection of the Task creation option, the task item is created in relation to the content item from which it was created and is displayed in a calendar. For example, the system identifies the “volunteer registration” task in the email and that the task needs to be completed tonight—relative to the day of creation or a date in the content item. The system creates the task and displays the task in the calendar application for completion tonight. In one example, the user is provided with an option whether to accept the task or to make any desired changes to the task such as for example, change the date and or time, reassigned the task, etc.; paragraph 0052).
The Sodhi-Herman combination describes features related to task management. Gruber is directed to a system and method for generating and processing task items. Therefore, they are deemed to be analogous as they both are directed towards solutions for task management and large language models. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the Sodhi-Herman combination with Gruber because the references are analogous art because they are both directed to solutions for task management and generative models, which falls within applicant’s field of endeavor (task management system), and because modifying the Sodhi-Herman combination to include Gruber’s feature for selecting, at the task management system, the set of supplemental tasks based on the list including the task, based on creation dates of tasks associated with the list, in the manner claimed, would serve the motivation of assisting a user in managing his/her tasks (Gruber at paragraph 0028); and further obvious because the claimed invention is merely a combination of old elements, and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable.
As per claim 3, the Sodhi-Herman-Gruber-Fowler combination teaches the method of claim 1. Although not explicitly taught by Sodhi, Herman in the analogous art of task management systems teaches wherein selecting, at the task management system, the set of supplemental tasks based on the list including the task and based on measures of similarity between an embedding determined for the task and embeddings determined for tasks associated with the list comprises: selecting tasks associated with the list including the task, created by the user, and created within a threshold amount of time from a time when the task management system received the task from the user as the set of supplemental tasks (paragraph 0014, discussing receiving a first user input directed to a digital assistant operating on the computer system; and in response to receiving the first user input: in accordance with a determination that the first user input corresponds to a request to initiate a plan, wherein the plan includes one or more tasks suggested by the digital assistant: displaying a first response graphical object corresponding to the plan, wherein the first response graphical object includes a representation of the one or more tasks suggested by the digital assistant; and adding an ongoing plan graphical object corresponding to the plan to a user interface that indicates one or more ongoing plans; paragraph 0296, discussing a context module that determines the context information relevant at a current time and/or current location of computer system. In some examples, the current time refers to the time when a request is made (e.g., by the user) for LLM unit to provide LLM output. In some examples, the context module scores different context information accessible to computer system based on current relevance and determines the relevant context information to include the context information having respective scores that each exceed a threshold. In some examples, context module determines the relevant context information based on analyzing current and/or historical user interactions associated with computer system; paragraph 0311, discussing that once the task(s) are determined, each task is assessed as to whether the task satisfies management criteria…In some examples, LLM model tuning is used to determine capabilities of computer system. Specifically, a task-specific dataset may be applied to LLM 910…The fine-tuned LLM may then be used to assess system capabilities. Tasks satisfying the management criteria are then be added to an ordered task list. Any tasks not satisfying the management criteria may be subject to additional follow-up. For instance, the user may be prompted to modify the task request and/or provide additional information in order to facilitate proper processing of the task (e.g., the user may be prompted “Did you mean BestNewsPage.com?”); paragraph 0312, discussing that each task of the ordered task list may include various task dependency criteria. In general, the task dependency criteria may indicate which (if any) tasks must be performed prior to a respective task being performed. For example, in order to perform the fourth task “(4) Compose message to John's contact identifier including link to identified headline story,” the second task “(2) Identify link to headline story on ‘BestNews.com’ within web browser application” must first be performed. Likewise, in order to perform the second task “(2) Identify link to headline story on ‘BestNews.com’ within web browser application” the first task “(1) Navigate to ‘BestNews.com’ using web browser application” must first be performed. Generally, tasks that are dependent on the performance of other tasks within the ordered task list may be referred to as synchronous tasks, whereas tasks that are not dependent on the performance of any other tasks within the ordered task list may be referred to as asynchronous tasks; paragraph 0417, discussing determining (e.g., using an AI process or a generative AI process), based on the plurality of tasks, an ordered task list associated with a set of task dependency criteria, wherein each task of the ordered task list includes a respective task dependency criteria (e.g., each task includes information on whether the task is dependent on the result of other tasks or is independent of the result of other tasks). In some examples, based on the parsed commands, a set of tasks for satisfying the commands is determined (e.g., using an AI process or a generative AI process). In some examples, a determination (e.g., using an AI process or a generative AI process) is made whether each task of the set of tasks is manageable; paragraph 0329).
The Sodhi-Herman combination does not explicitly teach selecting, at the task management system, the set of supplemental tasks based on the list including the task, based on creation dates of tasks associated with the list. Gruber in the analogous art of task generation systems teaches:
selecting, at the task management system, the set of supplemental tasks based on the list including the task, based on creation dates of tasks associated with the list (paragraph 0032, discussing that a suggestion engine is operable to determine whether a candidate task received from a tasks source should be suggested to the user as a task to perform at a given time and/or location. From all of the candidate task items that may be presented to the user at any given time, the suggestion engine filters those task items to a manageable subset based on the user's existing task items, prior acceptances/rejections of suggested task items, and the prior actions of the user. For example, if a user's calendar includes an event for an upcoming birthday, a suggested task is created that the person whose birthday is coming up should be called prior to that date; paragraph 0035, discussing that a preview generator is operable to generate previews for entities associated with a suggested task (or a selected task); paragraph 0048, discussing that upon selection of the task creation control, the “volunteer registration” task is created as a task item for “tonight”. In another aspect, the user selects the task creation option and identifies the objects, persons, and times relevant to that task; paragraph 0049, discussing that upon selection of the Task creation option, the task item is created in relation to the content item from which it was created and is displayed in a calendar. For example, the system identifies the “volunteer registration” task in the email and that the task needs to be completed tonight—relative to the day of creation or a date in the content item. The system creates the task and displays the task in the calendar application for completion tonight. In one example, the user is provided with an option whether to accept the task or to make any desired changes to the task such as for example, change the date and or time, reassigned the task, etc.; paragraph 0052).
The Sodhi-Herman combination describes features related to task management. Gruber is directed to a system and method for generating and processing task items. Therefore, they are deemed to be analogous as they both are directed towards solutions for task management and large language models. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the Sodhi-Herman combination with Gruber because the references are analogous art because they are both directed to solutions for task management and generative models, which falls within applicant’s field of endeavor (task management system), and because modifying the Sodhi-Herman combination to include Gruber’s feature for selecting, at the task management system, the set of supplemental tasks based on the list including the task, based on creation dates of tasks associated with the list, in the manner claimed, would serve the motivation of assisting a user in managing his/her tasks (Gruber at paragraph 0028); and further obvious because the claimed invention is merely a combination of old elements, and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable.
As per claim 5, the Sodhi-Herman-Gruber-Fowler combination teaches the method of claim 1. Although not explicitly taught by Sodhi, Herman in the analogous art of task management systems teaches wherein selecting, at the task management system, the set of supplemental tasks based on the list including the task, and based on measures of similarity between an embedding determined for the task and embeddings determined for tasks associated with the list comprises: generating the embedding for the task based on received information from the user describing the task (paragraph 0284, discussing that input data is received and provided to tokenization module which converts input data into a token and/or a series of tokens which can be processed by input embedding module into a format that is understood by foundation model. The tokenization module converts input data into a series of characters that has a specific semantic meaning to the foundation model; paragraph 0286, discussing that after the input data has been tokenized, the input data is provided to input an embedding module to convert the tokens to a vector representation that can be processed by the foundation model…The input embedding module converts the various data provided as an input into a format that foundation model can parse and process; paragraph 0305, discussing that the input embedding layer passes the vector representation to the LLM, which processes the vector representation in order to obtain LLM output. The LLM may include various neural network layers and other mechanisms, including one or more attention mechanisms, feed-forward layers, normalization, and residual connections to facilitate processing of the vector representation to obtain LLM output. Attention mechanisms serve to weight various aspects of the vector representation, which in turn allow the LLM to prioritize processing the most relevant portions of the embedding (e.g., particular words and/or phrases) and identify relationships between different portions of the embedding (e.g., dependencies between different tasks within a task list). Normalization layers and residual connections may assist the LLM with identifying main components of an input and generalizing concepts across different types of words, phrases, and context. Feed-forward layers transform an input into a new representation in order to identify patterns within the input embedding (e.g., focusing on general types of words or phrases when certain context is present). Moreover, the LLM may include multiple iterations of processing using the components described);
determining embeddings for additional tasks maintained by the task management system (paragraph 0287, discussing that when the foundation model is a large language model (LLM) the tokenization module converts input data into text which is then converted into a vector representation by the input embedding module that can be processed by the foundation model to determine a response to the input data as output or to determine a summary of input data as output; paragraph 0305, discussing that the input embedding layer passes the vector representation to the LLM, which processes the vector representation in order to obtain LLM output. The LLM may include various neural network layers and other mechanisms, including one or more attention mechanisms, feed-forward layers, normalization, and residual connections to facilitate processing of the vector representation to obtain LLM output. Attention mechanisms serve to weight various aspects of the vector representation, which in turn allow the LLM to prioritize processing the most relevant portions of the embedding (e.g., particular words and/or phrases) and identify relationships between different portions of the embedding (e.g., dependencies between different tasks within a task list). Normalization layers and residual connections may assist the LLM with identifying main components of an input and generalizing concepts across different types of words, phrases, and context. Feed-forward layers transform an input into a new representation in order to identify patterns within the input embedding (e.g., focusing on general types of words or phrases when certain context is present). Moreover, the LLM may include multiple iterations of processing using the components described); and
selecting additional tasks having embeddings with at least a threshold measure of similarity to the embedding for the task as the set of supplemental tasks (paragraph 0204, discussing that a comparison between the characteristic intensity and one or more thresholds is used to determine whether or not to perform one or more operations; paragraph 0246, discussing that the natural language processing module receives the candidate text representations…, and for each candidate representation, determines what nodes are implicated by the words in the candidate text representation. In some examples, if a word or phrase in the candidate text representation is found to be associated with one or more nodes in ontology 760, the word or phrase “triggers” or “activates” those nodes. Based on the quantity and/or relative importance of the activated nodes, the natural language processing module selects one of the actionable intents as the task that the user intended the digital assistant to perform. In some examples, the domain that has the most “triggered” nodes is selected. In some examples, the domain having the highest confidence value (e.g., based on the relative importance of its various triggered nodes) is selected. In some examples, the domain is selected based on a combination of the number and the importance of the triggered nodes. In some examples, additional factors are considered in selecting the node as well, such as whether the digital assistant has previously correctly interpreted a similar request from a user; paragraph 0307, discussing that the classification type included in the LLM output indicates whether a user request (e.g., represented by the input embedding) corresponds to a request to initiate a plan. For example, the LLM is trained to classify an input as a request to initiate a plan (a planning request) or as not a request to initiate a plan (a non-planning request). Generally, a plan includes, at least initially, one or more tasks suggested by the computer system for completion of the plan. Non-limiting examples of planning requests include a request to plan a vacation (e.g., “help me plan a trip to Tokyo”), a request for an exercise plan (e.g., “help me run a 6 minute mile”), a request for a diet plan (e.g., “help me eat healthier”), a request to plan an event (e.g., “help me plan Thanksgiving dinner”),…, and other requests to achieve a user-specified goal. The computer system interprets such planning requests as requests to generate respective suggestion(s) associated with the plan, e.g., suggestions for steps/tasks associated with planning a trip to Tokyo or suggestions for steps/tasks associated with running a mile in 6 minutes or less...Whether a particular request is classified as a planning request or as a non-planning request can depend on the particular implementation of the LLM. As described below with respect to FIGS. 12A-12Y, in some examples, computer system performs different actions responsive to a request depending on whether it is classified as a planning request or as a non-planning request. Accordingly, in some examples, planning requests and non-planning requests are distinguished by the differing actions computer system performs in response to receiving the respective request; paragraph 0308, discussing that the LLM output is then passed to task flow module which performs various functions to satisfy the commands. The task flow module may also determine that additional information is needed in order to satisfy the commands, such that the task flow module passes one or more output representations back to the input embedding layer. In some examples, the output from the task flow module may first be passed to the tokenization layer, or alternatively, may be passed directly to the LLM (e.g., depending on the format of the output from task flow module). The task flow module may cause one of more outputs to be provided at the end of processing or throughout processing; paragraph 0306).
Sodhi is directed towards automation of tasks using language model prompts. Herman is directed to systems and techniques for incorporating large language models into intelligent automated assistants. Therefore, they are deemed to be analogous as they both are directed towards solutions for task management and large language models. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine Sodhi with Herman because the references are analogous art because they are both directed to solutions for task management and generative models, which falls within applicant’s field of endeavor (task management system), and because modifying Sodhi to include Herman’s features for including wherein selecting, at the task management system, the set of supplemental tasks based on the list including the task, and based on measures of similarity between an embedding determined for the task and embeddings determined for tasks associated with the list comprises: generating the embedding for the task based on received information from the user describing the task; determining embeddings for additional tasks maintained by the task management system; and selecting additional tasks having embeddings with at least a threshold measure of similarity to the embedding for the task as the set of supplemental tasks, in the manner claimed, would serve the motivation of allowing more efficient and accurate management of user requested plans (Herman at paragraph 0018); and further obvious because the claimed invention is merely a combination of old elements, and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable.
The Sodhi-Herman combination does not explicitly teach selecting, at the task management system, the set of supplemental tasks based on the list including the task, based on creation dates of tasks associated with the list. Gruber in the analogous art of task generation systems teaches:
selecting, at the task management system, the set of supplemental tasks based on the list including the task, based on creation dates of tasks associated with the list (paragraph 0032, discussing that a suggestion engine is operable to determine whether a candidate task received from a tasks source should be suggested to the user as a task to perform at a given time and/or location. From all of the candidate task items that may be presented to the user at any given time, the suggestion engine filters those task items to a manageable subset based on the user's existing task items, prior acceptances/rejections of suggested task items, and the prior actions of the user. For example, if a user's calendar includes an event for an upcoming birthday, a suggested task is created that the person whose birthday is coming up should be called prior to that date; paragraph 0035, discussing that a preview generator is operable to generate previews for entities associated with a suggested task (or a selected task); paragraph 0048, discussing that upon selection of the task creation control, the “volunteer registration” task is created as a task item for “tonight”. In another aspect, the user selects the task creation option and identifies the objects, persons, and times relevant to that task; paragraph 0049, discussing that upon selection of the Task creation option, the task item is created in relation to the content item from which it was created and is displayed in a calendar. For example, the system identifies the “volunteer registration” task in the email and that the task needs to be completed tonight—relative to the day of creation or a date in the content item. The system creates the task and displays the task in the calendar application for completion tonight. In one example, the user is provided with an option whether to accept the task or to make any desired changes to the task such as for example, change the date and or time, reassigned the task, etc.; paragraph 0052).
The Sodhi-Herman combination describes features related to task management. Gruber is directed to a system and method for generating and processing task items. Therefore, they are deemed to be analogous as they both are directed towards solutions for task management and large language models. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the Sodhi-Herman combination with Gruber because the references are analogous art because they are both directed to solutions for task management and generative models, which falls within applicant’s field of endeavor (task management system), and because modifying the Sodhi-Herman combination to include Gruber’s feature for selecting, at the task management system, the set of supplemental tasks based on the list including the task, based on creation dates of tasks associated with the list, based on creation dates of tasks associated with the list, in the manner claimed, would serve the motivation of assisting a user in managing his/her tasks (Gruber at paragraph 0028); and further obvious because the claimed invention is merely a combination of old elements, and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable.
As per claim 6, the Sodhi-Herman-Gruber-Fowler combination teaches the method of claim 5. Although not explicitly taught by Sodhi, Herman in the analogous art of task management systems teaches wherein each additional task was created by the user (paragraph 0253, discussing that the task flow processing module is configured to receive the structured query (or queries) from the natural language processing module, complete the structured query, if necessary, and perform the actions required to “complete” the user's ultimate request. In some examples, the various procedures necessary to complete these tasks are provided in the task flow models. In some examples, the task flow models include procedures for obtaining additional information from the user and task flows for performing actions associated with the actionable intent; paragraph 0311, discussing that once the task(s) are determined, each task is assessed as to whether the task satisfies management criteria. Management criteria may generally correspond to criteria needed to determine whether the computer system can perform the task. Assessment of task management criteria may be performed in multiple ways...For example, the task “(1) Navigate to ‘BestNews.com’ using web browser application” is determined to be manageable given that the computer system has an application corresponding to a web browser. In some examples, LLM model tuning is used to determine capabilities of the computer system…Specifically, a task-specific dataset may be applied to the LLM , such that the LLM architecture and LLM parameters are optimized in order to handle requests directed to a specific domain or area of interest (e.g., trip planning, cooking, scholarly research, social media interactions, etc.). The fine-tuned LLM may then be used to assess system capabilities. Tasks satisfying the management criteria are then added to an ordered task list. Any tasks not satisfying the management criteria may be subject to additional follow-up. For instance, the user may be prompted to modify the task request and/or provide additional information in order to facilitate proper processing of the task; paragraph 0433, discussing receiving a user input on a displayed textual representation of a respective task, and in accordance with a determination that the respective task includes at least one of additional information or one or more sub-tasks, displaying at least the additional information and the one or more sub-tasks).
Sodhi is directed towards automation of tasks using language model prompts. Herman is directed to systems and techniques for incorporating large language models into intelligent automated assistants. Therefore, they are deemed to be analogous as they both are directed towards solutions for task management and large language models. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine Sodhi with Herman because the references are analogous art because they are both directed to solutions for task management and generative models, which falls within applicant’s field of endeavor (task management system), and because modifying Sodhi to include Herman’s feature for including wherein each additional task was created by the user, in the manner claimed, would serve the motivation of allowing more efficient and accurate management of user requested plans (Herman at paragraph 0018); and further obvious because the claimed invention is merely a combination of old elements, and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable.
As per claim 7, the Sodhi-Herman-Gruber-Fowler combination teaches the method of claim 6. Although not explicitly taught by Sodhi, Herman in the analogous art of task management systems teaches wherein each additional task is associated with the list including the task (paragraph 0253, discussing that the task flow processing module is configured to receive the structured query (or queries) from the natural language processing module, complete the structured query, if necessary, and perform the actions required to “complete” the user's ultimate request. In some examples, the various procedures necessary to complete these tasks are provided in the task flow models. In some examples, the task flow models include procedures for obtaining additional information from the user and task flows for performing actions associated with the actionable intent; paragraph 0314, discussing that as tasks within the ordered task list are performed, context information may be accumulated and used for optimized task performance. For example, the user may request a local dinner reservation for four specific people along with a request to send the reservation confirmation to the attendees. While performing the ordered task list, a determination is made that there are two locations for the respective restaurant that are generally reachable by the user (e.g., within 10 miles). Context information regarding each restaurant may then be gathered in order to determine whether additional follow-up with the user is needed. For example, the additional context information may include restaurant ratings, reservation availability, the user's preference regarding either location, and the like. To the extent a determination cannot be made as to which restaurant the user intends to dine at, an additional task can be created (e.g., “Provide prompt to user asking whether the user prefers to dine at Restaurant X or Restaurant Y). The additional task can then be inserted into the ordered task list at the appropriate slot based on any relevant task dependency criteria; paragraph 0315).
Sodhi is directed towards automation of tasks using language model prompts. Herman is directed to systems and techniques for incorporating large language models into intelligent automated assistants. Therefore, they are deemed to be analogous as they both are directed towards solutions for task management and large language models. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine Sodhi with Herman because the references are analogous art because they are both directed to solutions for task management and generative models, which falls within applicant’s field of endeavor (task management system), and because modifying Sodhi to include Herman’s feature for including wherein each additional task is associated with the list including the task, in the manner claimed, would serve the motivation of allowing more efficient and accurate management of user requested plans (Herman at paragraph 0018); and further obvious because the claimed invention is merely a combination of old elements, and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable.
As per claim 8, the Sodhi-Herman-Gruber-Fowler combination teaches the method of claim 7. Although not explicitly taught by Sodhi, Herman in the analogous art of task management systems teaches wherein each additional task was created within a threshold amount of time of a time when the task management system received the task (paragraph 0403, discussing that once the device generates a plan, the DA (digital assistant) enables the user to provide input (e.g., natural language input) to update the plan. Example updates to a plan correcting a step of the plan, providing more information about the plan, changing an objective and/or a step of the plan, adding a step to the plan, removing a step from the plan, or otherwise modifying the plan. The device indicates such updates to the plan via the graphical object for the plan and/or via the detailed user interface for the plan. In some examples, the device receives the input via selection of a natural language input entry field displayed in the detailed user interface for the plan. In some examples, the device receives the input while displaying a user interface different from the detailed user interface. As a specific example, the device receives input requesting “for my Tokyo trip plan, add the task of making a reservation at a good sushi place” and in response, the device updates the plan to include the additional task, e.g., by updating the detailed user interface to include a graphical element for making a reservation at a sushi restaurant. In some examples, based on a user input that provides more information about a plan, the device updates one or more steps of the plan based on the provided information. For example, based on user input “I'm taking my kids on the Tokyo trip,” the device updates the suggested activities in detailed user interface to include family friendly activities, e.g., by requesting LLM unit to generate suggestions for activities to do in Tokyo based on context information indicating that children are coming along; paragraph 0407, discussing that the device receives speech input (“let's pick this up later”). The device determines that speech input corresponds to a request to pause the task and in accordance with such determination, the device adds corresponding task object to the planning user interface. While the example shows that the request to pause the task is a speech input, in other examples, the device receives another type of input that is interpreted as a request to pause the task. For example, if the device receives an input that requests to dismiss the DA user interface while the task is incomplete (e.g., before device sets the reminder), the device interprets the input as a request to pause the task and thus adds the corresponding task object to the planning user interface; paragraph 0409, discussing that the corresponding task object allows the user to resume the paused task. For example, the device receives a user input corresponding to a selection of the corresponding task object and in response, displays a user interface associated with the paused task. The user interface enables the user to provide a set of inputs to complete the task, e.g., to specify timing information for the reminder and to specify the subject of the reminder. Adding task objects that correspond to paused tasks to the planning user interface provides a convenient and efficient way for the user to return to paused tasks, thereby allowing the user to use the device more quickly and efficiently).
Sodhi is directed towards automation of tasks using language model prompts. Herman is directed to systems and techniques for incorporating large language models into intelligent automated assistants. Therefore, they are deemed to be analogous as they both are directed towards solutions for task management and large language models. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine Sodhi with Herman because the references are analogous art because they are both directed to solutions for task management and generative models, which falls within applicant’s field of endeavor (task management system), and because modifying Sodhi to include Herman’s feature for including wherein each additional task was created within a threshold amount of time of a time when the task management system received the task, in the manner claimed, would serve the motivation of allowing more efficient and accurate management of user requested plans (Herman at paragraph 0018); and further obvious because the claimed invention is merely a combination of old elements, and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable.
As per claim 9, the Sodhi-Herman-Gruber-Fowler combination teaches the method of claim 5. Although not explicitly taught by Sodhi, Herman in the analogous art of task management systems teaches wherein the embedding for the task is determined based on a name of the task and a description of the task (paragraph 0245, discussing that each node in ontology 760 is associated with a set of words and/or phrases that are relevant to the property or actionable intent represented by the node. The respective set of words and/or phrases associated with each node are the so-called “vocabulary” associated with the node. The respective set of words and/or phrases associated with each node are stored in vocabulary index 744 in association with the property or actionable intent represented by the node. For example, returning to FIG. 7B, the vocabulary associated with the node for the property of “restaurant” includes words such as “food,” “drinks,” “cuisine,” “hungry,” “eat,” “pizza,” “fast food,” “meal,” and so on. For another example, the vocabulary associated with the node for the actionable intent of “initiate a phone call” includes words and phrases such as “call,” “phone,” “dial,” “ring,” “call this number,” “make a call to,” and so on; paragraph 0287, discussing that when the foundation model is a large language model (LLM) the tokenization module converts input data into text which is then converted into a vector representation by the input embedding module that can be processed by the foundation model to determine a response to the input data as output or to determine a summary of input data as output; paragraph 0246, discussing that the natural language processing module receives the candidate text representations (e.g., text string(s) or token sequence(s)) from the STT (speech-to-text) processing module, and for each candidate representation, determines what nodes are implicated by the words in the candidate text representation. In some examples, if a word or phrase in the candidate text representation is found to be associated with one or more nodes in ontology 760, the word or phrase “triggers” or “activates” those nodes. Based on the quantity and/or relative importance of the activated nodes, the natural language processing module selects one of the actionable intents as the task that the user intended the digital assistant to perform. In some examples, the domain that has the most “triggered” nodes is selected. In some examples, the domain having the highest confidence value (e.g., based on the relative importance of its various triggered nodes) is selected. In some examples, the domain is selected based on a combination of the number and the importance of the triggered nodes. In some examples, additional factors are considered in selecting the node as well, such as whether the digital assistant has previously correctly interpreted a similar request from a user; paragraph 0305, discussing that the input embedding layer passes the vector representation to the LLM, which processes the vector representation in order to obtain LLM output. The LLM may include various neural network layers and other mechanisms, including one or more attention mechanisms, feed-forward layers, normalization, and residual connections to facilitate processing of the vector representation to obtain LLM output. Attention mechanisms serve to weight various aspects of the vector representation, which in turn allow the LLM to prioritize processing the most relevant portions of the embedding (e.g., particular words and/or phrases) and identify relationships between different portions of the embedding (e.g., dependencies between different tasks within a task list). Normalization layers and residual connections may assist the LLM with identifying main components of an input and generalizing concepts across different types of words, phrases, and context. Feed-forward layers transform an input into a new representation in order to identify patterns within the input embedding (e.g., focusing on general types of words or phrases when certain context is present). Moreover, the LLM may include multiple iterations of processing using the components described).
Sodhi is directed towards automation of tasks using language model prompts. Herman is directed to systems and techniques for incorporating large language models into intelligent automated assistants. Therefore, they are deemed to be analogous as they both are directed towards solutions for task management and large language models. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine Sodhi with Herman because the references are analogous art because they are both directed to solutions for task management and generative models, which falls within applicant’s field of endeavor (task management system), and because modifying Sodhi to include Herman’s feature for including wherein the embedding for the task is determined based on a name of the task and a description of the task, in the manner claimed, would serve the motivation of allowing more efficient and accurate management of user requested plans (Herman at paragraph 0018); and further obvious because the claimed invention is merely a combination of old elements, and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable.
As per claim 10, the Sodhi-Herman-Gruber-Fowler combination teaches the method of claim 1. Although not explicitly taught by Sodhi, Herman in the analogous art of task management systems teaches modifying the selected candidate subtask in response to the task management system receiving an input from the user (paragraph 0307, discussing that the classification type included in the LLM output indicates whether a user request (e.g., represented by the input embedding) corresponds to a request to initiate a plan. For example, the LLM is trained to classify an input as a request to initiate a plan (a planning request) or as not a request to initiate a plan. Generally, a plan includes, at least initially, one or more tasks suggested by the computer system for completion of the plan. Non-limiting examples of planning requests include a request to plan a vacation (e.g., “help me plan a trip to Tokyo”), a request for an exercise plan (e.g., “help me run a 6 minute mile”), a request for a diet plan (e.g., “help me eat healthier”), a request to plan an event (e.g., “help me plan Thanksgiving dinner”),…, and other requests to achieve a user-specified goal. The computer system interprets such planning requests as requests to generate respective suggestion(s) associated with the plan, e.g., suggestions for steps/tasks associated with planning a trip to Tokyo or suggestions for steps/tasks associated with running a mile in 6 minutes or less...Whether a particular request is classified as a planning request or as a non-planning request can depend on the particular implementation of the LLM. As described below with respect to FIGS. 12A-12Y, in some examples, computer system performs different actions responsive to a request depending on whether it is classified as a planning request or as a non-planning request. Accordingly, in some examples, planning requests and non-planning requests are distinguished by the differing actions computer system performs in response to receiving the respective request; paragraph 0308, discussing that the LLM output is then passed to task flow module which performs various functions to satisfy the commands. The task flow module may also determine that additional information is needed in order to satisfy the commands, such that the task flow module passes one or more output representations back to the input embedding layer. In some examples, the output from the task flow module may first be passed to the tokenization layer, or alternatively, may be passed directly to the LLM (e.g., depending on the format of the output from task flow module). The task flow module may cause one of more outputs to be provided at the end of processing or throughout processing; paragraph 0311, discussing that once the task(s) are determined, each task is assessed as to whether the task satisfies management criteria…In some examples, LLM model tuning is used to determine capabilities of computer system. Specifically, a task-specific dataset may be applied to LLM 910, such that the LLM architecture and LLM parameters are optimized in order to handle requests directed to a specific domain or area of interest. The fine-tuned LLM may then be used to assess system capabilities. Tasks satisfying the management criteria are then be added to an ordered task list…For instance, the user may be prompted to modify the task request and/or provide additional information in order to facilitate proper processing of the task; paragraph 0417, discussing determining (e.g., using an AI process or a generative AI process), based on the plurality of tasks, an ordered task list associated with a set of task dependency criteria, wherein each task of the ordered task list includes a respective task dependency criteria. In some examples, based on the parsed commands, a set of tasks for satisfying the commands is determined (e.g., using an AI process or a generative AI process). In some examples, a determination (e.g., using an AI process or a generative AI process) is made whether each task of the set of tasks is manageable; paragraphs 0246, 0312).
Sodhi is directed towards automation of tasks using language model prompts. Herman is directed to systems and techniques for incorporating large language models into intelligent automated assistants. Therefore, they are deemed to be analogous as they both are directed towards solutions for task management and large language models. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine Sodhi with Herman because the references are analogous art because they are both directed to solutions for task management and generative models, which falls within applicant’s field of endeavor (task management system), and because modifying Sodhi to include Herman’s feature for including modifying the selected candidate subtask in response to the task management system receiving an input from the user, in the manner claimed, would serve the motivation of allowing more efficient and accurate management of user requested plans (Herman at paragraph 0018); and further obvious because the claimed invention is merely a combination of old elements, and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable.
Claim 20 recites substantially similar limitations that stand rejected via the art citations and rationale applied to claim 1, as discussed above. Further, as per claim 20 the Sodhi-Herman-Gruber-Fowler combination teaches a method performed at a task management system comprising a processor and a non-transitory computer readable medium (Sodhi, paragraph 0135, discussing that the methods and systems described may be deployed in part or in whole through a machine that executes computer software, program codes, and/or instructions on a processor. “Processor” as used herein is meant to include at least one processor and unless context clearly indicates otherwise, the plural and the singular should be understood to be interchangeable. Any aspects of the present disclosure may be implemented as a computer-implemented method on the machine, as a system or apparatus as part of or in relation to the machine, or as a computer program product embodied in a computer readable medium executing on one or more of the machines.). Although not explicitly taught by Sodhi, Herman in the analogous art of task management systems teaches selecting, at the task management system, a set of supplemental tasks that each have one or more specific attributes (paragraph 0311, discussing that once the task(s) are determined, each task is assessed as to whether the task satisfies management criteria…In some examples, LLM model tuning is used to determine capabilities of computer system. Specifically, a task-specific dataset may be applied to LLM 910, such that the LLM architecture and LLM parameters are optimized in order to handle requests directed to a specific domain or area of interest. The fine-tuned LLM may then be used to assess system capabilities. Tasks satisfying the management criteria are then be added to an ordered task list. Any tasks not satisfying the management criteria may be subject to additional follow-up. For instance, the user may be prompted to modify the task request and/or provide additional information in order to facilitate proper processing of the task (e.g., the user may be prompted “Did you mean BestNewsPage.com?”); paragraph 0312, discussing that each task of the ordered task list may include various task dependency criteria. In general, the task dependency criteria may indicate which (if any) tasks must be performed prior to a respective task being performed. For example, in order to perform the fourth task “(4) Compose message to John's contact identifier including link to identified headline story,” the second task “(2) Identify link to headline story on ‘BestNews.com’ within web browser application” must first be performed. Likewise, in order to perform the second task “(2) Identify link to headline story on ‘BestNews.com’ within web browser application” the first task “(1) Navigate to ‘BestNews.com’ using web browser application” must first be performed. Generally, tasks that are dependent on the performance of other tasks within the ordered task list may be referred to as synchronous tasks, whereas tasks that are not dependent on the performance of any other tasks within the ordered task list may be referred to as asynchronous tasks; paragraph 0417, discussing determining (e.g., using an AI process or a generative AI process), based on the plurality of tasks, an ordered task list associated with a set of task dependency criteria, wherein each task of the ordered task list includes a respective task dependency criteria (e.g., each task includes information on whether the task is dependent on the result of other tasks or is independent of the result of other tasks). In some examples, based on the parsed commands, a set of tasks for satisfying the commands is determined (e.g., using an AI process or a generative AI process). In some examples, a determination (e.g., using an AI process or a generative AI process) is made whether each task of the set of tasks is manageable).
Sodhi is directed towards automation of tasks using language model prompts. Herman is directed to systems and techniques for incorporating large language models into intelligent automated assistants. Therefore, they are deemed to be analogous as they both are directed towards solutions for task management and large language models. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine Sodhi with Herman because the references are analogous art because they are both directed to solutions for task management and generative models, which falls within applicant’s field of endeavor (task management system), and because modifying Sodhi to include Herman’s feature for including selecting, at the task management system, a set of supplemental tasks that each have one or more specific attributes, in the manner claimed, would serve the motivation of allowing more efficient and accurate management of user requested plans (Herman at paragraph 0018); and further obvious because the claimed invention is merely a combination of old elements, and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable.
Claim 11 recites substantially similar limitations that stand rejected via the art citations and rationale applied to claim 1, as discussed above. Further, as per claim 11 the Sodhi-Herman-Gruber-Fowler combination teaches a computer program product comprising a non-transitory computer readable storage medium having instructions encoded thereon that, when executed by a processor, cause the processor to perform steps (Sodhi, paragraph 0135, discussing that the methods and systems described may be deployed in part or in whole through a machine that executes computer software, program codes, and/or instructions on a processor. “Processor” as used herein is meant to include at least one processor and unless context clearly indicates otherwise, the plural and the singular should be understood to be interchangeable. Any aspects of the present disclosure may be implemented as a computer-implemented method on the machine, as a system or apparatus as part of or in relation to the machine, or as a computer program product embodied in a computer readable medium executing on one or more of the machines).
Claim 12 recites substantially similar limitations that stand rejected via the art citations and rationale applied to claim 2, as discussed above.
Claim 13 recites substantially similar limitations that stand rejected via the art citations and rationale applied to claim 3, as discussed above.
Claim 15 recites substantially similar limitations that stand rejected via the art citations and rationale applied to claim 5, as discussed above.
Claim 16 recites substantially similar limitations that stand rejected via the art citations and rationale applied to claim 6, as discussed above.
Claim 17 recites substantially similar limitations that stand rejected via the art citations and rationale applied to claim 7, as discussed above.
Claim 18 recites substantially similar limitations that stand rejected via the art citations and rationale applied to claim 8, as discussed above.
Claim 19 recites substantially similar limitations that stand rejected via the art citations and rationale applied to claim 10, as discussed above.
Allowable Subject Matter
19. Claims 4 and 14 are objected to as being dependent upon rejected base claims, but would be allowable if rewritten in independent form including all of the limitations of the base claim and any intervening claims. Claim 4 recites “The method of claim 1, wherein selecting, at the task management system, the set of supplemental tasks based on the list including the task, based on creation dates of tasks associated with the list, and based on measures of similarity between an embedding determined for the task and embeddings determined for tasks associated with the list comprises: selecting additional tasks associated with the list including the task and created by the user; generating a ranking of the additional tasks based on times when the additional tasks were created relative to a time when the task management system received the task from the user, where additional tasks more recently created relative to the time when the task management system received the task have higher positions in the ranking; and selecting additional tasks having at least a threshold position in the ranking as the set of supplemental tasks.” The closest prior art, Sar et al., Patent No.: US 10,831,348 B1, describes “ranking one or more of the associated class members for the task component class identifier, where the ranking of a given associated class member of the associated class members may be based on frequency of occurrence of the given associated class member in the task completion indicators” (col. 1, lines 50-54) and that” one or more of the associated class members for the task component class identifier may be ranked. The ranking may be based on frequency of occurrence of the given associated class member in the task completion indicators. For example, the task component class identifier may be an action class identifier, and the ranking system may rank the plurality of potential action members. Also, for example, the task component class identifier may be a class of entities identifier, and the ranking system may rank the plurality of potential entity members. In some implementations the content database may store data related to frequency with which actions and/or entities are utilized to complete a task” (col. 12, lines 53-65). While Sar et al., involves a system that, give a task of a class, accesses members (components) of that class and ranks them by frequency of completion. Then suggests or selects some of those based on ranking. The ranking is based on frequency, not on a creatin time relative to when the user submitted the task, as required by the claim. Also, the “additional tasks” being selected are “task components” or entities rather than tasks previously created by the user. Sar does not provide an indication that the selection threshold is based on certain recency relative to the main task receipt. None of the prior art describe the process of wherein selecting, at the task management system, the set of supplemental tasks based on the list including the task, based on creation dates of tasks associated with the list, and based on measures of similarity between an embedding determined for the task and embeddings determined for tasks associated with the list comprises: selecting additional tasks associated with the list including the task and created by the user; generating a ranking of the additional tasks based on times when the additional tasks were created relative to a time when the task management system received the task from the user, where additional tasks more recently created relative to the time when the task management system received the task have higher positions in the ranking; and selecting additional tasks having at least a threshold position in the ranking as the set of supplemental tasks, as recited in claim 4 (and similarly recited in clam 14). While the cited prior art describe the concept of ranking tasks, the cited prior art does not detail the specific ranking operations involving combining creation time recency relative to task receipt with thresholding to select supplementals tasks, specifically to be included in a prompt or subtask generation chain. Sodhi, Herman, Sar, and the other prior art of record does not teach the claim limitations directed to specific techniques for wherein selecting, at the task management system, the set of supplemental tasks based on the list including the task, based on creation dates of tasks associated with the list, and based on measures of similarity between an embedding determined for the task and embeddings determined for tasks associated with the list comprises: selecting additional tasks associated with the list including the task and created by the user; generating a ranking of the additional tasks based on times when the additional tasks were created relative to a time when the task management system received the task from the user, where additional tasks more recently created relative to the time when the task management system received the task have higher positions in the ranking; and selecting additional tasks having at least a threshold position in the ranking as the set of supplemental tasks ” as recited in claim 4 (and similarly recited in claim 14), thus rendering claims 4 and 14 as allowable over the prior art. Claims 4 and 14 are not allowable, however, because claims 4 and 14 remain rejected under 35 U.S.C. 101. Furthermore, even if the §101 rejection of claims 4 and 14 is overcome, claims 4 and 14 would be objected to as being dependents upon rejected base claim (claim 1 and 11, respectively). Claim 4 is objected to as being dependent upon rejected base claim 1, but would be allowable if rewritten in independent form including all of the limitations of their respective base claims and any intervening claims, as well as resolving all other outstanding rejections of this claim. Claim 14 is objected to as being dependent upon rejected base claim 11, but would be allowable if rewritten in independent form including all of the limitations of their respective base claims and any intervening claims, as well as resolving all other outstanding rejections of this claim.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Pollock et al., Pub. No.: US 2025/0356313 A1 – describes multi-agent task management guided by generative artificial intelligence.
Maschmeyer et al., Pub. No.: US 2024/0311546 A1 – describes methods and systems for prompting large language model to generate formatted output.
Garigliotti, Darío, and Krisztian Balog. "Generating query suggestions to support task-based search." Proceedings of the 40th International ACM SIGIR Conference on Research and Development in Information Retrieval. 2017 – describes a tool for supporting task-based search.
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any extension fee pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to DARLENE GARCIA-GUERRA whose telephone number is (571) 270-3339. The examiner can normally be reached M-F 7:30a.m.-5:00p.m. EST.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Brian M. Epstein can be reached on (571) 270-5389. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/Darlene Garcia-Guerra/
Primary Examiner, Art Unit 3625