Detailed Action
Status of Application
The following is a Final Office Action. In response to Examiner's communication on 01/27/2026, Applicant on 06/29/2026, amended Claims 1, 13, 17-20, cancelled Claims 8 and 16, and added new Claims 21-24. Claims 1-7, 9-15, 17-24 are now pending in this application and have been rejected below.
Response to Amendment
Applicant’s amendments are sufficient to overcome the 35 USC 112(b) rejections set forth in the previous action. Therefore, these rejections have been withdrawn accordingly.
Applicants’ amendments render moot the 35 USC 102 rejections set forth in the previous action in view of new and updated grounds for rejection necessitated by Applicants’ amendments. Therefore, these rejections are withdrawn in view of the new grounds for rejection necessitated by Applicants’ amendments, as set forth below.
Applicants’ amendments are insufficient to overcome the 35 USC 103 rejections set forth in the previous action. Therefore, these rejections have been updated to address the amendments and are maintained below.
Response to Arguments – 35 USC § 102(a)(1)
Applicant's arguments with respect to the rejection of Claims under 35 USC 102(a)(1) have been considered but are moot in light of new grounds of rejections necessitated by applicant’s amendments. Examiner respectfully notes updated rejections below under 35 USC 103.
Response to Arguments – 35 USC § 103
Applicant's arguments with respect to the rejection of Claims under 35 USC 103 have been considered but are not found to be persuasive.
Applicant notes the legal guidelines for a proper rejection under 35 USC 103. Examiner notes it is somewhat unclear as to what specific limitations Applicant disagrees with as being taught by the prior art, or which particular combination of references is supported by an unclear rationale. In view of this, Examiner respectfully notes the rejections as updated to address amendments below.
Claim Rejections - 35 USC § 103
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 (i.e., changing from AIA to pre-AIA ) 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.
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, 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.
Claims 1-2, 4-5, 8, 11-13, 16, 18-20, 22 are rejected under 35 U.S.C. 103 as being unpatentable over Zhang(US 20180052664 A1) in view of O'Malia(US 20220036153 A1).
Claims 1, 13
A tangible, non-transitory, machine-readable medium storing instructions that when executed by one or more processors effectuate operations, comprising: receiving, from a computing device, a request for an agent, the request including a set of one or more tasks to be automatically completed by the agent;
In [0007], " In one example, a method implemented on a computer having at least one processor, a storage, and a communication platform for developing a virtual agent is disclosed...The set of modules are then integrated in the order to generate the virtual agent which, when deployed, performs actions corresponding to the set of modules in the order". In [0045], "The virtual agent development engine 170 in this example may develop a customized virtual agent for a developer via a bot design programming interface provided to the developer...The virtual agent development engine 170 may also store the customized tasks into the customized task database 139, which can provide previously generated tasks as a template for future task generation or customization during virtual agent development". We understand these stored templates to be particular commands that can execute a task.
Additionally, we have support for the live update of tasks via user requests derived from usage in [0044], "During the conversation between the virtual agent and the user, the virtual agent can analyze dialog states of the dialog and manage real-time tasks related to the dialog, based on data stored in various databases, e.g. a knowledge database 134, a publisher database 136, and a customized task database 139".
obtaining, from a database associated with the computing device, one or more preconfigured commands for performing at least some of the set of one or more tasks;
In [0045], "The virtual agent development engine 170 in this example may develop a customized virtual agent for a developer via a bot design programming interface provided to the developer...The virtual agent development engine 170 may also store the customized tasks into the customized task database 139, which can provide previously generated tasks as a template for future task generation or customization during virtual agent development". We understand these stored templates to be particular commands that can execute a task.
Additionally, we have support for the live update of tasks and derivative commands via user requests derived from usage in [0044], "During the conversation between the virtual agent and the user, the virtual agent can analyze dialog states of the dialog and manage real-time tasks related to the dialog, based on data stored in various databases, e.g. a knowledge database 134, a publisher database 136, and a customized task database 139. The virtual agent may also perform product/service recommendation to the user based on a user database 132. In one embodiment, when the virtual agent determines that the user's intent has changed or the user is unsatisfied with the current dialog, the virtual agent may redirect the user to a different agent based on a virtual agent database 1".
determining that the database does not include a preconfigured command configured to perform a given task of the set of one or more tasks and, in response, generating, using an artificial-intelligence model, a generated command for performing the given task
In [0090], "Upon receiving the instruction for integrating, the visual input based program integrator 1012 in this example may integrate the modules obtained from the virtual agent module determiner 1006. For each of the modules, the visual input based program integrator 1012 may retrieve program source code for the module from the virtual agent program database 1014. For modules that have parameters customized based on inputs of the developer, the visual input based program integrator 1012 may modify the obtained source codes for the module based on the customized parameters. In one embodiment, the visual input based program integrator 1012 may invoke the machine learning engine 1016 to further modify the codes based on machine learning".
generating, using the one or more preconfigured commands,
a virtual agent configured to perform the one or more preconfigured commands and the generated command; and providing an indication that the agent is configured to perform the set of one or more tasks and available for deployment.
In [0096], "The modified program codes are integrated at 1118 to generate a customized virtual agent. The customized virtual agent is stored and sent at 1120 to the developer". The delivery of the generated agent serves as an indication of readiness for deployment. We also have intermediate development status updates in [0095], "One or more virtual agent modules are determined at 1108 based on the inputs. The development status of the virtual agent is stored or updated at 1110".
Zhang does not expressly disclose the remaining limitations.
However, O’Malia teaches:
generating, using an artificial-intelligence model, a generated command for performing the given task based at least in part on documentation uploaded by a user that describes a process for performing the given task ;
The ULLM inherently shapes agent behavior by means of its LLM functionality, in [0056], "The Al Agent Controller 102 may provide a system for the translation or transformation of visual and/or other data from the Al Agent environment into a format which may be optimized for and processed by the ULLM 114 to provide outputs which may, through presentation to the Al Agent 112, shape Al Agent action selection in a favorable manner, and which may enable the Al Agent 112 to generalize its abilities to more diverse environments than it has seen in the past".
In [0025], “The Al Agent Controller may be queried by the Al Agent via our method, and may by virtue of our system's bi-directional information conversion capability acquire or be provided with information regarding the environment and environment state of the Al Agent and components within it, including but not limited to, semantic and other labels, user manuals, human writing or voice content regarding the environment. This information regarding the environment may be exchanged or acquired via any other means and may include its relevance to other scenarios, games, environments, news, media, writing, and other recorded or streamed media”.
Zhang discloses a system for developing artificial agents. O'Malia discloses a system meant to enhance artificial agent performance. Each reference discloses means of utilizing and interfacing with AI agents. Extending the AI-guided agent management as recorded in O'Malia to the system of Zhang is applicable as they share the AI agent field of endeavor.
It would have been obvious to one having ordinary skill in the art at the effective filling date of the invention to apply the AI guided agent management of O'Malia and apply that to the system of Zhang. Motivation to do so comes from the fact that the claim is plainly directed to the predictable result of combining known items in the prior art, with the expected benefit that adopting the AI guided agent management of O'Malia found in [0023] of O'Malia, "The provided methods and/or systems enabling an Al Agent to benefit from the encoded mental models and or knowledge in the Al Agent Controller/ULLM may enable the Al Agent to incorporate, or make use of, internal or external knowledge bases, or facts encoded in the Al agent's past training, or other fact-related techniques as part of the Al Agent's capability set".
Claim 13 is rejected as disclosing substantially similar limitations as Claim 1.
Claim 2
Zhang teaches:
The medium of claim 1, the operations further comprising: receiving a request for changing performance of the agent; and updating at least one of the set of one or more tasks based upon the received request. In [0065], "FIG. 4 depicts an exemplary high level system diagram of a dynamic dialog state analyzer 210 in a service virtual agent, e.g. the service virtual agent 1 142 in FIG. 2, according to an embodiment of the present teaching. The dynamic dialog state analyzer 210 can keep track of the dialog state of the conversation with the user and the user's intent based on continuously received user input. The dialog state and user intent are also continuously updated based on the new input from the user. As shown in FIG. 4, the dynamic dialog state analyzer 210 comprises a parser 402, one or more natural language models 404, a dictionary 406, a dialog state generator 408, and a dialog log recorder 410". In [0068], "For example, upon receiving all related answers of the user extracted from the user input regarding a selling product, the dialog state generator 408 may retrieve a dialog state from the dialog log database 212 and update the dialog state to indicate that the user is ready to buy the product, and it is time to provide payment method or platform to the user".
Claim 4
Zhang teaches:
The medium of claim 1, the operations further comprising generating an executable set of computer code for one or more commands in accordance with a determination that the database does not include one or more preconfigured commands configured to perform at least one of the one or more tasks.
With respect to generating source code in response to need for customized program execution, in [0090], "Upon receiving the instruction for integrating, the visual input based program integrator 1012 in this example may integrate the modules obtained from the virtual agent module determiner 1006. For each of the modules, the visual input based program integrator 1012 may retrieve program source code for the module from the virtual agent program database 1014. For modules that have parameters customized based on inputs of the developer, the visual input based program integrator 1012 may modify the obtained source codes for the module based on the customized parameters. In one embodiment, the visual input based program integrator 1012 may invoke the machine learning engine 1016 to further modify the codes based on machine learning".
Claim 5
As to Claim 5, Zhang teaches all the limitations of Claim 1 as discussed above.
Zhang does not expressly disclose the remaining limitations.
However, O’Malia teaches:
The medium of claim 1, the operations further comprising generating a text document including text describing each of the one or more tasks of the set of one or more tasks to be performed by the agent.
See [0056] regarding the intended purpose of ULLM output, "The Al Agent Controller 102 may provide a system for the translation or transformation of visual and/or other data from the Al Agent environment into a format which may be optimized for and processed by the ULLM 114 to provide outputs which may, through presentation to the Al Agent 112, shape Al Agent action selection in a favorable manner, and which may enable the Al Agent 112 to generalize its abilities to more diverse environments than it has seen in the past". Specifying that its output is in text, in [0069], "The Al Agent Controller 102 may provide (208) the output text to the Al Agent 112. The Al Agent 112 may map semantics provided in the output text to actions". See [0068] for an example of ULLM output producing text instructions.
It would have been obvious to one having ordinary skill in the art at
the effective filling date of the invention to apply the Al guided agent
management of O'Malia and apply that to the system of Zhang. Motivation
to do so comes from the same rationale as outlined above with respect to Claim 1.
Claims 11, 18
Zhang teaches:
The medium of claim 1, the operations further comprising generating an updated version of the command by providing inputs of human interactions with computer software into the AI model.
As part of our command updating occurs from live interaction with the user as outlined above in our rejection of Claim 1, in [0037], "More specifically, based on machine learning and AI technique, the disclosed system can learn how to strategically ask user questions, present intermediate candidates to the users based on historical human-human or human-machine or machine-machine conversation data, together with human or machine action data that involves calling third party applications, services or databases…The disclosed system can use the knowledge base and historical conversations for recommending high quality response messages for future conversation". Claim 18 is rejected as disclosing substantially similar limitations as Claim 11.
Claim 12, 19
Zhang teaches:
The medium of claim 1, the operations further comprising training the AI model by providing human user input to generate updated versions of commands.
As part of our command updating occurs from live interaction with the user as outlined above in our rejection of Claim 1, in [0037], "More specifically, based on machine learning and AI technique, the disclosed system can learn how to strategically ask user questions, present intermediate candidates to the users based on historical human-human or human-machine or machine-machine conversation data, together with human or machine action data that involves calling third party applications, services or databases". Given that this human user data is salient to guiding actions of the agent, we consider this to be relevant to updating commands. Claim 19 is rejected as disclosing substantially similar limitations as Claim 12.
Claim 20
Zhang teaches:
The method of claim 13, further comprising steps for providing artificial compute agents.
In [0096], "The modified program codes are integrated at 1118 to generate a customized virtual agent. The customized virtual agent is stored and sent at 1120 to the developer". The delivery of the generated agent serves as an indication of readiness for deployment. We also have intermediate development status updates in [0095], "One or more virtual agent modules are determined at 1108 based on the inputs. The development status of the virtual agent is stored or updated at 1110".
Claim 22
As to Claim 22, Zhang teaches:
The method of claim 13, further comprising configuring the Al model with reinforcement learning with human feedback.
In [0037], “The present teaching has disclosed both statistical learning and template based approach as well as deep learning models (e.g… a reinforcement learning model…The disclosed system is also capable of using those information as well as users' implicit feedback signals (such as clicks and conversions) when interacting with our recommendation results to more effectively learn users' interests, persuade them for certain conversions, collect their explicit feedback (such as rating), as well as actively solicit additional sophisticated user feedback such as their suggestions for future product/service improvement”..
Claims 3, 9-10, 17 are rejected under 35 U.S.C. 103 as being unpatentable over Zhang(US 20180052664 A1) in view of O'Malia(US 20220036153 A1) in further view of Ravichandran(US 10963754 B1).
Claim 3
As to Claim 3, Zhang combined with O’Malia teaches all the limitations of Claim 1 as discussed above.
Zhang combined with O’Malia does not expressly disclose the remaining limitations.
However, Ravichandran teaches:
The medium of claim 1, wherein the request for the agent further comprises a sample output of the set of one or more tasks, the sample output including one or more text, audio, image, and video.
In Col 7 Lines 8-33, "FIG. 6 illustrates embodiments of a few-shot image classification service graphical user interface (GUI). In particular, this is a part of an overall GUI that allows a user to upload and classify images and/or train a model of his/her choosing with a small (on a relative scale) data set images. While this description is geared toward image classification, the GUI may be modified to fit the needs of other classifications (such as video, audio, etc.). In some implementations, the user interacts with this service (such as the few-shot classification service 506) via intermediate networks and interfaces such as those detailed herein. In this illustration, GUI 600 allows the user to request several different types of actions be performed by the few-shot image classification service. In particular, through this interface, a user may upload one or more images through an add image block 603 and then provide a label for the one or more images using an input box 602. Images may be added using a drop box 605 and/or by providing an image location 607 for the image (such as a URL or a pointer to a location in storage of a provider network). Once an image is “added,” in some embodiments, the user uploads the image via an upload button 609. In other embodiments, the upload occurs automatically. In this GUI, the user may also cause a training of the model via a request using train model input box 601".
Zhang combined with O’Malia discloses a system for developing artificial agents. Ravichandran discloses a system meant to leverage few shot learning in the application of AI models. Each reference discloses means of utilizing and interfacing with artificial intelligence. Extending the few shot usage of Ravichandran is applicable to the system of Zhang combined with O’Malia as they share the field of endeavor of artificial intelligence.
It would have been obvious to one having ordinary skill in the art at the effective filling date of the invention to leverage the few shot prompting as taught in Ravichandran and apply that to the system of Zhang combined with O’Malia. Motivation to do so comes from the fact that the claim is plainly directed to the predictable result of combining known items in the prior art, with the expected benefit that adopting said few shot learning would enable users to aid users in calibrating agents to perform tasks that may not have extensive representation in the training data.
Claim 9
As to Claim 9, Zhang combined with O’Malia teaches all the limitations of Claim 8 as discussed above.
Zhang teaches:
The medium of claim 1, the operations further comprising generating an updated version of the command
As part of our command updating occurs from live interaction with the user as outlined above in our rejection of Claim 1, in [0065], "FIG. 4 depicts an exemplary high level system diagram of a dynamic dialog state analyzer 210 in a service virtual agent, e.g. the service virtual agent 1 142 in FIG. 2, according to an embodiment of the present teaching. The dynamic dialog state analyzer 210 can keep track of the dialog state of the conversation with the user and the user's intent based on continuously received user input. The dialog state and user intent are also continuously updated based on the new input from the user. As shown in FIG. 4, the dynamic dialog state analyzer 210 comprises a parser 402, one or more natural language models 404, a dictionary 406, a dialog state generator 408, and a dialog log recorder 410". In [0068], "For example, upon receiving all related answers of the user extracted from the user input regarding a selling product, the dialog state generator 408 may retrieve a dialog state from the dialog log database 212 and update the dialog state to indicate that the user is ready to buy the product, and it is time to provide payment method or platform to the user".
Zhang combined with O’Malia does not expressly disclose the remaining limitations.
However, Ravichandran teaches:
by providing a video input into the AI model.
We consider the few shot prompting that can be used to shape the functionality of the AI models to disclose this limitation, in Col 7 Lines 8-15, "While this description is geared toward image classification, the GUI may be modified to fit the needs of other classifications (such as video, audio, etc.). In some implementations, the user interacts with this service (such as the few-shot classification service 506) via intermediate networks and interfaces such as those detailed herein".
It would have been obvious to one having ordinary skill in the art at the effective filling date of the invention to leverage the few shot prompting as taught in Ravichandran and apply that to the system of Zhang combined with O’Malia. Motivation to do so comes from the same rationale as outlined above with respect to Claim 3.
Claim 10
As to Claim 10, Zhang combined with O’Malia teaches all the limitations of Claim 8 as discussed above.
Zhang teaches:
The medium of claim 1, the operations further comprising generating an updated version of the command
As part of our command updating occurs from live interaction with the user as outlined above in our rejection of Claim 1, in [0065], "FIG. 4 depicts an exemplary high level system diagram of a dynamic dialog state analyzer 210 in a service virtual agent, e.g. the service virtual agent 1 142 in FIG. 2, according to an embodiment of the present teaching. The dynamic dialog state analyzer 210 can keep track of the dialog state of the conversation with the user and the user's intent based on continuously received user input. The dialog state and user intent are also continuously updated based on the new input from the user. As shown in FIG. 4, the dynamic dialog state analyzer 210 comprises a parser 402, one or more natural language models 404, a dictionary 406, a dialog state generator 408, and a dialog log recorder 410". In [0068], "For example, upon receiving all related answers of the user extracted from the user input regarding a selling product, the dialog state generator 408 may retrieve a dialog state from the dialog log database 212 and update the dialog state to indicate that the user is ready to buy the product, and it is time to provide payment method or platform to the user".
Zhang combined with O’Malia does not expressly disclose the remaining limitations.
However, Ravichandran teaches:
by providing an audio input into the AI model.
We consider the few shot prompting that can be used to shape the functionality of the AI models to disclose this limitation, in Col 7 Lines 8-15, "While this description is geared toward image classification, the GUI may be modified to fit the needs of other classifications (such as video, audio, etc.). In some implementations, the user interacts with this service (such as the few-shot classification service 506) via intermediate networks and interfaces such as those detailed herein".
It would have been obvious to one having ordinary skill in the art at the effective filling date of the invention to leverage the few shot prompting as taught in Ravichandran and apply that to the system of Zhang combined with O’Malia. Motivation to do so comes from the same rationale as outlined above with respect to Claim 3.
Claim 17
As to Claim 17, Zhang combined with O’Malia teaches all the limitations of Claim 13 as discussed above.
Zhang teaches:
The medium of claim 13, further comprising generating an updated version of the command
As part of our command updating occurs from live interaction with the user as outlined above in our rejection of Claim 1, in [0065], "FIG. 4 depicts an exemplary high level system diagram of a dynamic dialog state analyzer 210 in a service virtual agent, e.g. the service virtual agent 1 142 in FIG. 2, according to an embodiment of the present teaching. The dynamic dialog state analyzer 210 can keep track of the dialog state of the conversation with the user and the user's intent based on continuously received user input. The dialog state and user intent are also continuously updated based on the new input from the user. As shown in FIG. 4, the dynamic dialog state analyzer 210 comprises a parser 402, one or more natural language models 404, a dictionary 406, a dialog state generator 408, and a dialog log recorder 410". In [0068], "For example, upon receiving all related answers of the user extracted from the user input regarding a selling product, the dialog state generator 408 may retrieve a dialog state from the dialog log database 212 and update the dialog state to indicate that the user is ready to buy the product, and it is time to provide payment method or platform to the user".
Zhang combined with O’Malia does not expressly disclose the remaining limitations.
However, Ravichandran teaches:
by providing a video input or an audio input into the AI model.
We consider the few shot prompting that can be used to shape the functionality of the AI models to disclose this limitation, in Col 7 Lines 8-15, "While this description is geared toward image classification, the GUI may be modified to fit the needs of other classifications (such as video, audio, etc.). In some implementations, the user interacts with this service (such as the few-shot classification service 506) via intermediate networks and interfaces such as those detailed herein".
It would have been obvious to one having ordinary skill in the art at the effective filling date of the invention to leverage the few shot prompting as taught in Ravichandran and apply that to the system of Zhang combined with O’Malia. Motivation to do so comes from the same rationale as outlined above with respect to Claim 3.
Claim 6-7, 14-15, 23 are rejected under 35 U.S.C. 103 as being unpatentable over Zhang(US 20180052664 A1) in view of in view of O'Malia(US 20220036153 A1) in further view of Davidson(US 20220289537 A1).
Claims 6, 14
As to Claim 6, Zhang combined with O’Malia teaches all the limitations of Claim 1 as discussed above.
Zhang combined with O’Malia does not expressly disclose the remaining limitations.However, Davidson teaches:
The medium of claim 1, the operations further comprising assigning a score associated with an automation level for each of the one or more tasks.
We understand this automation level to correspond to a prediction of agent capability to autonomously perform a task, in [0014], "For each of at least a subset of the predicted actions for a task, the robot agent 102 uses the learned model and various signals to predict whether the robot agent will fail to perform the action in accordance with one or more policies. Such signals may include an indication of the quality of the action (e.g., a qualitative or quantitative evaluation of one or both the utility of the action and the risks of the action)(block 507), a qualitative or quantitative evaluation of the probability of failure in performing the action (block 508), and the like. If the resulting failure predictor indicates that the robot agent 102 is predicted to succeed at performing the task, then the robot agent 102 operates in an unguided mode to perform the task without seeking guidance from any of the guidance sources 110".
Davidson discloses a system for coordinating robot agents. Zhang combined with O’Malia discloses a system meant to create and deploy AI agents. Each reference discloses means of governing agentic operatives. Extending the failure prediction as recorded in Davidson to the system of Zhang combined with O’Malia is applicable as they both pertain to managing agents.
It would have been obvious to one having ordinary skill in the art at the effective filling date of the invention to apply the task success prediction of Davidson and apply that to the system of Zhang combined with O’Malia. Motivation to do so comes from the fact that the claim is plainly directed to the predictable result of combining known items in the prior art, with the expected benefit that adopting said prediction would enable intelligent routing and/or assistance to human operatives. See [0072] of Zhang, Davidson improves upon this functionality by allowing predictions to be made as to whether rerouting is needed.
Claim 14 is rejected as disclosing substantially similar limitations as Claim 6.
Claim 7, 15
As to Claim 7, Zhang combined with O’Malia teaches all the limitations of Claim 1 as discussed above.
Zhang combined with O’Malia does not expressly disclose the remaining limitations.However, Davidson teaches:
The medium of claim 1, the operations further comprising providing, for each task of the set of one or more tasks, a prediction score associated with a probability that completion of the task requires human intervention
In [0028], "The ultimate decision on whether performance of the predicted action is likely to fail may include a straightforward thresholding approach (e.g., by comparing each signal to a corresponding threshold, or by generating a final failure score and comparing this single score to a corresponding threshold". In [0014], "Conversely, if the resulting failure predictor indicates that the robot agent 102 is predicted to fail in performing the task, then the robot agent 102 operates in a guided mode in which the robot agent 102 seeks guidance input from one or more of the guidance sources 110 by transmitting a guidance request 114 to the agent coordination subsystem 106.". Guidance sources encompass human operators in [0012]. "The guidance sources 110 operate as “experts” or “teachers” for the robot agents 102, and may be implemented as human operators providing feedback or other guidance through a computer interface".
It would have been obvious to one having ordinary skill in the art at the effective filling date of the invention to apply the task success prediction of Davidson and apply that to the system of Zhang combined with O’Malia. Motivation to do so comes from the same rationale as outlined above with respect to Claim 6.
Claim 15 is rejected as disclosing substantially similar limitations as Claim 7.
Claim 23
As to Claim 23, Zhang combined with O’Malia teaches all the limitations of Claim 13 as discussed above.
Zhang teaches:
The method of claim 13, wherein the one or more preconfigured commands and the generated command each comprise data that causes functionality to be invoked… the generated command, and wherein performance of the generated command by the virtual agent causes at least a portion of the given task to be performed.
With respect to generating source code in response to need for customized program execution, in [0090], "Upon receiving the instruction for integrating, the visual input based program integrator 1012 in this example may integrate the modules obtained from the virtual agent module determiner 1006. For each of the modules, the visual input based program integrator 1012 may retrieve program source code for the module from the virtual agent program database 1014. For modules that have parameters customized based on inputs of the developer, the visual input based program integrator 1012 may modify the obtained source codes for the module based on the customized parameters. In one embodiment, the visual input based program integrator 1012 may invoke the machine learning engine 1016 to further modify the codes based on machine learning". In [0096], "The modified program codes are integrated at 1118 to generate a customized virtual agent. The customized virtual agent is stored and sent at 1120 to the developer".
Zhang does not expressly disclose the remaining limitations. However, O’Malia teaches:
wherein generating the generated command comprises analyzing, using the artificial-intelligence model, the documentation uploaded by the user that describes the process for performing the given task and generating, based on contents of the documentation,
In [0025], “The Al Agent Controller may be queried by the Al Agent via our method, and may by virtue of our system's bi-directional information conversion capability acquire or be provided with information regarding the environment and environment state of the Al Agent and components within it, including but not limited to, semantic and other labels, user manuals, human writing or voice content regarding the environment. This information regarding the environment may be exchanged or acquired via any other means and may include its relevance to other scenarios, games, environments, news, media, writing, and other recorded or streamed media”. Note that the con troller is based on large language artificial intelligence models in [0021], “The systems and methods may, through bi-directional information conversion between Al Agent environment and Al Agent Controller/ULLM, enable the Al Agent to use information which may be encoded in the ULLM”.
It would have been obvious to one having ordinary skill in the art at
the effective filling date of the invention to apply the Al guided agent
management of O'Malia and apply that to the system of Zhang. Motivation
to do so comes from the same rationale as outlined above with respect to Claim 1.
Zhang combined with O’Malia does not expressly disclose the remaining limitations.
However, Davidson teaches:
the data comprising one or more arguments of a function call or an application- programming-interface call,
In [0017], “The software stack 208 includes one or more software programs, that is, sets of executable instructions that, when executed by one or more processors of the compute hardware subsystem 206, manipulate components of the compute hardware subsystem 206, the electro-mechanical subsystem 202, and the sensor subsystem 204 to perform various operations in furtherance of the techniques and processes described herein. To this end, the software programs may utilize libraries, application programming interfaces (APIs), and other similar ancillary software tools in performing these operations”.
It would have been obvious to one having ordinary skill in the art at the effective filling date of the invention to apply the task success prediction of Davidson and apply that to the system of Zhang combined with O’Malia. Motivation to do so comes from the same rationale as outlined above with respect to Claim 6.
Claim 21 is rejected under 35 U.S.C. 103 as being unpatentable over Zhang(US 20180052664 A1) in view of O'Malia(US 20220036153 A1) in further view of Prompt Programming for Large Language Models: Beyond the Few-Shot Paradigm (Reynolds et. al, 2021).
Claim 21
As to Claim 21, Zhang teaches:
The method of claim 13, wherein generating the virtual agent comprises
In [0045], "The virtual agent development engine 170 in this example may develop a customized virtual agent for a developer via a bot design programming interface provided to the developer”.
Zhang combined with O’Malia does not expressly disclose the remaining limitations.
However, Reynolds teaches:
configuring a preamble injected before a prompt to the Al model.
See Figure 1 at the bottom of Page 3 of the attached NPL. The few-shot prompting technique of supplying examples prior to prompt entry is an example of such a preamble injection.
Zhang combined with O’Malia discloses a system for developing artificial agents, with orchestration facilitated by an LLM. Reynolds discloses a method of enhanced prompting to LLMs. Each reference discloses means of leveraging LLMs. Extending the prompting techniques as recorded in Reynolds to the system of Zhang combined with O’Malia is applicable as they share the problem of optimally leveraging LLMs to process contextual information.
It would have been obvious to one having ordinary skill in the art at the effective filling date of the invention to apply the prompting methodology of Reynolds and apply that to the system of Zhang combined with O’Malia. Motivation to do so comes from the fact that the claim is plainly directed to the predictable result of combining known items in the prior art, with the expected benefit that adopting the prompting techniques of Reynolds would improve LLM performance. See Table 1 for demonstrated gains in performance on Page 3 of the attached Reynolds NPL.
Claim 24 is rejected under 35 U.S.C. 103 as being unpatentable over Zhang(US 20180052664 A1) in view of O'Malia(US 20220036153 A1) in further view of Davidson(US 20220289537 A1) in further view of Quick Reference Guide: Creating and Submitting an Expense Report (2021).
Claim 24
As to Claim 24, Zhang combined with O’Malia teaches all the limitations of Claim 13 as discussed above.
Zhang teaches:
The method of claim 13, … wherein the one or more preconfigured commands and the generated command each comprise…and wherein the method further comprises generating an updated version of the generated command by providing inputs of human interactions with computer software into the artificial-intelligence model and configuring the virtual agent to perform the updated version of the generated command.
With respect to generating source code in response to need for customized program execution, in [0090], "Upon receiving the instruction for integrating, the visual input based program integrator 1012 in this example may integrate the modules obtained from the virtual agent module determiner 1006. For each of the modules, the visual input based program integrator 1012 may retrieve program source code for the module from the virtual agent program database 1014. For modules that have parameters customized based on inputs of the developer, the visual input based program integrator 1012 may modify the obtained source codes for the module based on the customized parameters. In one embodiment, the visual input based program integrator 1012 may invoke the machine learning engine 1016 to further modify the codes based on machine learning". As part of our command updating occurs from live interaction with the user as outlined above in our rejection of Claim 1, in [0037], "More specifically, based on machine learning and AI technique, the disclosed system can learn how to strategically ask user questions, present intermediate candidates to the users based on historical human-human or human-machine or machine-machine conversation data, together with human or machine action data that involves calling third party applications, services or databases…The disclosed system can use the knowledge base and historical conversations for recommending high quality response messages for future conversation".
Zhang does not expressly disclose the remaining limitations. However, O’Malia teaches:
wherein the documentation uploaded by the user comprises…wherein the artificial-intelligence model analyzes the … and contents of the … are applied to the virtual agent to train the virtual agent in the process…
In [0025], “The Al Agent Controller may be queried by the Al Agent via our method, and may by virtue of our system's bi-directional information conversion capability acquire or be provided with information regarding the environment and environment state of the Al Agent and components within it, including but not limited to, semantic and other labels, user manuals, human writing or voice content regarding the environment. This information regarding the environment may be exchanged or acquired via any other means and may include its relevance to other scenarios, games, environments, news, media, writing, and other recorded or streamed media”. Note that the con troller is based on large language artificial intelligence models in [0021], “The systems and methods may, through bi-directional information conversion between Al Agent environment and Al Agent Controller/ULLM, enable the Al Agent to use information which may be encoded in the ULLM”.
It would have been obvious to one having ordinary skill in the art at
the effective filling date of the invention to apply the Al guided agent
management of O'Malia and apply that to the system of Zhang. Motivation
to do so comes from the same rationale as outlined above with respect to Claim 1.
Zhang combined with O’Malia does not expressly disclose the remaining limitations.
However, Davidson teaches:
the data that causes functionality to be invoked in a form of one or more arguments of a function call or an application- programming-interface call,
In [0017], “The software stack 208 includes one or more software programs, that is, sets of executable instructions that, when executed by one or more processors of the compute hardware subsystem 206, manipulate components of the compute hardware subsystem 206, the electro-mechanical subsystem 202, and the sensor subsystem 204 to perform various operations in furtherance of the techniques and processes described herein. To this end, the software programs may utilize libraries, application programming interfaces (APIs), and other similar ancillary software tools in performing these operations”.
It would have been obvious to one having ordinary skill in the art at the effective filling date of the invention to apply the task success prediction of Davidson and apply that to the system of Zhang combined with O’Malia. Motivation to do so comes from the same rationale as outlined above with respect to Claim 6.
Zhang combined with O’Malia and Davidson does not expressly disclose the remaining limitations.
However, Quick Reference Guide teaches:
wherein the given task comprises submitting an expense report…comprises a slide deck explaining a process for submitting the expense report…slide deck and contents of the slide deck…in the process for submitting the expense reports
See Pages 1-2 of attached NPL: Quick Reference Guide: Creating and Submitting an Expense Report for an example of such slide deck (the constituent pages) documentation.
Zhang combined with O’Malia and Davidson discloses a system for developing artificial agents, with orchestration informed by user manuals and input. Quick Reference Guide illustrates a means of compiling an expense report with slides. Extending the particular expense report methodology as recorded in Quick Reference Guide to the system of Zhang combined with O’Malia and Davidson is applicable as such documentation is reasonably pertinent to the problem of instructing agents, as facilitated by the ingestion of user documentation, that Zhang combined with O’Malia and Davidson addresses. A reference in a field different from that of applicant's endeavor may be reasonably pertinent if it is one which, because of the matter with which it deals, logically would have commended itself to an inventor's attention in considering his or her invention as a whole (MPEP 2141.01). The prior art of record provides common essential elements, even though the prior art applied is not concerned with providing such a slide deck to AI agents but rather the general usage of documentation to serve as a user manual, solving the pertinent problem of providing context to an AI agent.
It would have been obvious to one having ordinary skill in the art at the effective filling date of the invention to apply the instructions of Quick Reference Guide and apply that to the system of Zhang combined with O’Malia and Davidson. Motivation to do so comes from the fact that the usage of Quick Reference Guide is a simple substitution for the documentation supported in [0025] of O’Malia, and the combination is derived from the predictable result of leveraging a particular form of documentation to said functionality in O’Malia.
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). 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 nonprovisional extension fee (37 CFR 1.17(a)) 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 THEODORE L XIE whose telephone number is (571)272-7102. The examiner can normally be reached M-F 9-5.
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, Rutao Wu can be reached at 571-272-6045. 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.
/THEODORE XIE/Examiner, Art Unit 3623
/WILLIAM S BROCKINGTON III/Primary Examiner, Art Unit 3623
Prosecution Insights