DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Status of Claims
This is a final office action in response to the amendment filed 03 March 2026. Claims 1, 11, and 20 have been amended. Claims 1-20 remain pending and have been examined.
Response to Amendment
Applicant’s amendment to claims 1, 11, and 20 have been entered.
Applicants’ amendment is insufficient to overcome the pending 35 U.S.C. 101 rejection. The rejection remains pending and is updated below, as necessitated by amendment.
Applicant’s amendment is insufficient to overcome the 35 U.S.C. 103 rejection. The rejection remains pending and is updated below, as necessitated by amendment.
Response to Arguments
Applicant’s arguments regarding the 35 U.S.C. 103 rejection have been fully considered, but is moot because the arguments do not apply to the combination of references used in the updated prior art rejection, detailed below.
Applicant’s arguments regarding the 35 U.S.C. 101 rejection have been fully considered, but are not persuasive. Applicant asserts that the claims do not recite a judicial exception because the system executes a sequence of action s to complete a portion of a task by invoking API function calls in a manner that is rooted in computing technology, specifically the field of generating application-specific workflows for tasks using a machine -learned model to generate a set of output tokens. Applicant further asserts that amended claim 1 explicitly integrates using generative AI (large language model-LLM) into the practical application of generating application-specific workflow, wherein the LLM-generated workflow includes a sequence of at least two function calls or API calls, where the output data from the first call is input into the second call, such that the workflow is automatically executed by a computer process. Applicant lastly asserts that the claims are an improvement in the technical filed of generating application-specific workflows for tasks because the online system solves the technical problem of significant coordination in retrieving information (e.g. through API calls) for complex and ambiguous tasks. Examiner respectfully disagrees.
Per paragraph [0017] of the specification the claims are directed to using a large language model to generate workflows for a picker and present items included in a customer’s order in a collection interface, as well as instructions that related to the collection of items in the order, presenting the location of each item in the retailer location, and may specify a sequence in which the picker should collect the items. Therefore, the claim limitations are directed to an abstract idea of data gathering and analysis steps for managing the workflow of a picker performing picker services, and fall within the certain methods of organizing human activity grouping of abstract concepts. The claims additionally fall with in the mental processes grouping of abstract concepts because a picker could mentally or through use of pen and paper make observations, judgements, and evaluations regarding how to fulfill a customer order and the sequence of steps that should be performed to complete the task.
The use of the LLM to generate an output in the form of a workflow or prompt response that is displayed for user decision making and action, without significantly more. The claims generative machine learning model ( LLM) and API are field of use applications of the related technology and tools used to implement the abstract idea of generating a task workflow. The claim limitations are not directed to an improvement to the underlying technologies, but to use of the technology to generate an output for presentation to a user. Further, the claimed first and second API calls are claimed in a manner that describes how an API call is used to present the results of the data analysis steps. The limitation for “executing the sequence of actions to complete at least a portion of the task,” is broadly and generically claimed to include the mere presentation of a query response or workflow recommendation as the output of the data analysis steps, without significantly more. See Specification at paragraph [0065-0067]. As a result, the 35 U.S.C. 101 rejection is proper, maintained, and updated below as necessitated by the amendment herein.
Claim Rejections - 35 USC § 101
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.
Claims 1-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea of collecting, analyzing, and outputting data for generating workflows, without significantly more. Independent claim 1 recites a process, independent claim 11 recites a product, and independent claim 20 recites a device for receiving orders from users via a web application and fulfilling the order with one or more retailer stores using one or more API calls, and providing results to a client device. Independent claim 1, 11, and 20 recite substantially similar limitations and are directed to an abstract idea of collecting and analyzing data, and presenting certain results of the collection and analysis for completion of a task based on user input.
Taking independent claim 1 as representative, amended claim 1 recites the following limitations:
receiving, from a client device, a description of a task;
responsive to receiving the description of the task, identifying a list of candidate actions from a database;
generating a prompt for a generative machine-learned model, the prompt including the description of the task, a request for a plan for the task, and the list of candidate actions and attributes related to the list of candidate actions;
providing the prompt to a model serving system deployed with the machine-learned model for execution;
encoding input data from the prompt as a set of input tokens using the model serving system, wherein the set of input tokens are encoded as embeddings in latent space; applying parameters of the machine-learned model to the set of input tokens to generate a set of output tokens, wherein the machine-learned model is configured as a transformer architecture including one or more attention layers;
obtaining a response from the set of output tokens to extract a plan including at least one workflow from the response, wherein the workflow includes a sequence of actions associated with the description of the task, wherein the sequence of actions comprises at least a first action and a second action, the first action comprising a first function call or application programming (API) call, a first node identifier, and a first set of arguments for input to the first function call or API call, the second action comprising a second function call or API call, a second node identifier, and a second set of arguments for input to the second function call or API call, wherein the second set of arguments includes at least output data from the first function call or API call;
executing the sequence of actions to complete at least a portion of the task, wherein executing the sequence of actions comprises: invoking the first function call or API call with the first set of arguments identified from the response, and invoking the second function call or API call with the second set of arguments identified from the response; and
providing results of the execution to the client device.
Under Step 1 of the eligibility analysis, independent claims 1, 11, and 20 recite at least one step or act, including receiving a description of a task. Thus the claims fall within one of the statutory categories.
Under Step 2A, Prong One these limitations recite certain methods of organizing human activity related to performing commercial interactions and/or fundamental economic principles because claim 1 recites extracting a workflow plan for fulfilling customer orders. (See Specification at paragraphs [0012-0021]). Per paragraph [0021-22] the order fulfilment “picker is a single person who collects items for an order from a retailer location and delivers the order to the delivery location for the order … the description primarily refers to pickers as humans.” The limitations further fall within the certain methods of organizing human activity grouping of abstract concepts because the results provided to the client device are workflow instructions provided to a human form completion. The claims additionally fall within the mental processes grouping of abstract concepts because a picker could mentally or through use of pen and paper make observations, judgements, and evaluations regarding how to fulfill a customer order and the sequence of steps that should be performed to complete the task. Accordingly, under Step 2A, Prong One, claim 1 recites an abstract idea.
Under Step 2A, Prong Two, the judicial exception of claim 1 is not integrated into a practical application. In particular, the claim recites a database, machine learning model, function call/or API, and computing system and memory of claims 11 and 20 to perform the recited steps. These elements are recited at a high level of generality (i.e., as generic computer components performing generic computer/ data processing functions) and amount to no more than mere instructions to apply the exception using generic computer components. For Example, Applicant’s Specification at paragraph [0094] states: “Any of the steps, operations, or processes described herein may be performed or implemented with one or more hardware or software modules, alone or in combination with other devices. In some embodiments, a software module is implemented with a computer program product comprising one or more computer-readable media storing computer program code or instructions, which can be executed by a computer processor for performing any or all of the steps, operations, or processes described.” The claims application programming interface (API) is used for its intended purpose and is construed as a field of use application of API technology. Further, the claimed first and second API calls are claimed in a manner that describes how an API call is used to present the results of the data analysis steps. The limitation for “executing the sequence of actions to complete at least a portion of the task,” is broadly and generically claimed to include the mere presentation of a query response or workflow recommendation as the output of the data analysis steps, without significantly more. See Specification at paragraph [0065-0067]. Adding generic computer components to perform generic functions, such as data gathering, performing calculations, and outputting a result would not transform the claim into eligible subject matter. See MPEP 2106.05(h). Additionally, the claims do not reflect an improvement in the functioning of a computer, or an improvement to other technology or technical filed (See MPEP 2106.05(a)); do not effect a transformation or reduction of a particular article to a different state or thing (See MPEP 2106.05(c)); and do not apply or use the abstract idea in some other meaningful way beyond generally linking the use of the abstract idea to a particular technological environment or filed of use (See MPEP 2106.05(e)).
The presence of machine learning algorithm as an additional element and computer limitations do not necessarily make the claim rooted in computer technology. As drafted the claimed generative machine learning model that received input via a prompt to output a workflow plan or prompt response for an order fulfillment task does not improve the functioning of the computing device, it improves the determination of a sequence of steps required to fulfil the order based on known data. As such, the claim requirement amounts to mere instructions to implement the abstract idea on a computer, and, therefore, is not sufficient to make the claim patent eligible. The recited machine learning function processes data to generate an output and is not an improvement to machine learning technology or the functioning of a computing device, and hence does not result in a practical application. Accordingly, the additional elements do not integrate the abstract idea into a practical application because they do not impose any meaningful limits on practicing the abstract idea.
Under Step 2B the claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. Because the Specification describes the additional elements in general terms, without describing the particulars, we conclude the claim limitations may be broadly but reasonably construed as reciting conventional computer components and techniques, particularly in light of Applicant’s Specification, as cited above. As discussed above with respect to the integration of the abstract idea into a practical application, the additional elements of a processor and storage device amount to no more than mere instructions to apply the exception using a generic computer component which cannot provide an inventive concept. See MPEP 2106.05.
Dependent claims 2-10 and 12-19 include the abstract ideas of the independent claims. The limitations of the dependent claims merely narrow the method of organizing human activity by identifying how the data is transmitted and processed, how the output is analyzed using at least one checkpoint, that the output plan includes a second workflow including another sequence of actions, determining risk related to the plan by obtaining a loss function, and describing attributes of the processed data. The limitations of the dependent claims are not integrated into a practical application because none of the additional elements set forth any limitations that meaningfully limit the abstract idea implementation. There are no additional elements that transform the claim into a patent eligible idea by amounting to significantly more. The analysis above applies to all statutory categories of invention. Accordingly independent claims 11 and 20 and the claims that depend therefrom are rejected as ineligible for patenting under 35 U.S.C. 101 based upon the same analysis applied to claim 1 above. Therefore claims 1- 20 are ineligible under 35 U.S.C. 101.
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.
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 non-obviousness.
Claims 1-7, 9, 11-17, and 19-20 are rejected under 35 U.S.C. 103 as being unpatentable over Koller et al. (US 2023/0380623), in view of Saxena (US 2024/0330579) and in further view of Kim et al. (US 2022/0222667).
Regarding Amended Claim 1, Koller et al. discloses a method comprising: receiving, from a client device, a description of a task; responsive to receiving the description of the task, identifying a list of candidate actions from a database; (… a method for preparing multiple beverages can include receiving multiple drink orders, adding the multiple drink orders on a queue, receiving a selection of a plurality of drink orders of the multiple drink orders on the queue, building recipes for each of the selected plurality of drink orders on the queue, and sending build steps from the recipes for each of the selected plurality of drink orders to corresponding substations. … To dynamically generate the sequence execution of the machine-readable instructions by the one or more processors can further cause the one or more processors to access a machine learning model for dynamically generating the sequence, wherein the machine learning model is configured to generate the sequence based on at least one of customer data or store data. Koller et al. [para. 0009-0015]. … the beverage production system allows for human input in deciding which beverages to create and which steps to take, which can allow beverages to be made quickly and also accommodate the complexity of the customizations. Koller et al. [para. 0048-0049]. … The computing system 3102 may include one or more internal and/or external data sources (for example, data sources. Koller et al. [para. 0146-0147]);
generating a prompt for a generative machine-learned model, the prompt including the description of the task, a request for a plan for the task, and the list of candidate actions and attributes related to the list of candidate actions; providing the prompt to a model serving system deployed with the machine-learned model for execution; ( The system can enable a user to select (e.g., choose) a sequence of the one or more sequences and may build collections of build instructions based on the selection (e.g., received via the dashboard). … the system can enable a user to customize a sequence of the one or more beverages and/or food items (e.g., by defining the sequence based on received user input) and/or can generate a customized sequence based on user input. … The system may include and/or utilize artificial intelligence to build the one or more sequences. In some cases, the system may utilize a machine learning model (e.g., a supervised, unsupervised, reinforcement, etc. machine learning model) to build the one or more sequences. The system may provide data associated with one or more orders, customer data associated with the one or more orders, store data, user data, ingredient data, machine data, or any other data associated with the one or more orders to the machine learning model as an input. Based on the provided input, the machine learning model may output one or more sequences. Koller et al. [para. 0037-0043, 0060-0065]);
Koller et al. fails to explicitly disclose encoding input data from the prompt as a set of input tokens using the model serving system, wherein the set of input tokens are encoded as embeddings in latent space; applying parameters of the machine-learned model to the set of input tokens to generate a set of output tokens, wherein the machine-learned model is configured as a transformer architecture including one or more attention layers. Saxena discloses these limitations. ( Input to a language model (whether transformer-based or otherwise) typically is in the form of natural language as may be parsed into tokens. … Each token 556 in the token sequence is converted into an embedding vector 560 (also referred to simply as an embedding). … another trained ML model may be used to convert the token 556 into an embedding 560 in a way that encodes additional information into the embedding 560 (e.g., a trained ML model may encode positional information about the position of the token 556 in the text sequence into the embedding 560). … The encoder 552 serves to encode the embeddings 560 into feature vectors 562 that represent the latent features of the embeddings 560. …the decoder 554 is designed to map the features represented by the feature vectors 562 into meaningful output. … Each output token 564 may be fed back as input to the decoder 554 in order to generate the next output token 564. By feeding back the generated output and applying self-attention, the decoder 554 is able to generate a sequence of output tokens 564 that has sequential meaning. … A computing system may generate a prompt that is provided as input to the LLM via its API. As described above, the prompt may optionally be processed or pre-processed into a token sequence prior to being provided as input to the LLM via its API. Saxena [para. 0065-0075, 0080]). It would have been obvious to one of ordinary skill in the art of data analysis via generative AI and automated workflow sequencing and communication protocols before the effective filing date of the claimed invention to modify the data analysis steps of Koller et al. to include input data from the prompt as a set of input tokens using the model serving system, wherein the set of input tokens are encoded as embeddings in latent space; applying parameters of the machine-learned model to the set of input tokens to generate a set of output tokens, wherein the machine-learned model is configured as a transformer architecture including one or more attention layers as disclosed by Saxena to enable the LLM to better generate output according to the desired output (Saxena [para. 0075]) in a manner that would have yielded predictable results at the relevant time.
obtaining a response… to extract a plan including at least one workflow from the response, wherein the workflow includes a sequence of actions associated with the description of the task, ( In response to a selected sequence of orders and/or a selected order, the system can output the selected sequence of orders and/or the selected order (e.g., via a display). Koller et al. [para. 0043; Fig. 2]);
While Koller et al. discloses the display of the sequence of actions of the workflow and communication across communication and device channels ( The display or dashboard can act as a user interface. Koller et al. [para. 0084]. … a display device provides for the presentation of GUIs as application software data. Koller et al. [para. 0136-0147]), Koller et al. and Saxena combined fail to explicitly disclose obtaining a response from the set of output tokens to extract a plan including at least one workflow from the response, wherein the workflow includes a sequence of actions associated with the description of the task, wherein the sequence of actions comprises at least a first action and a second action, the first action comprising a first function call or application programming (API) call, a first node identifier, and a first set of arguments for input to the first function call or API call, the second action comprising a second function call or API call, a second node identifier, and a second set of arguments for input to the second function call or API call. Saxena discloses these limitations. The encoder 552 serves to encode the embeddings 560 into feature vectors 562 that represent the latent features of the embeddings 560. …the decoder 554 is designed to map the features represented by the feature vectors 562 into meaningful output. … Each output token 564 may be fed back as input to the decoder 554 in order to generate the next output token 564. By feeding back the generated output and applying self-attention, the decoder 554 is able to generate a sequence of output tokens 564 that has sequential meaning. … A computing system may generate a prompt that is provided as input to the LLM via its API. As described above, the prompt may optionally be processed or pre-processed into a token sequence prior to being provided as input to the LLM via its API. Saxena [para. 0065-0075, 0080]). It would have been obvious to one of ordinary skill in the art of data analysis via generative AI and automated workflow sequencing and communication protocols before the effective filing date of the claimed invention to modify the data analysis steps of Koller et al. to include obtaining a response from the set of output tokens to extract a plan including at least one workflow from the response, wherein the workflow includes a sequence of actions associated with the description of the task, wherein the sequence of actions comprises at least a first action and a second action, the first action comprising a first function call or application programming (API) call, a first node identifier, and a first set of arguments for input to the first function call or API call, the second action comprising a second function call or API call, a second node identifier, and a second set of arguments for input to the second function call or API call as disclosed by Saxena to enable the LLM to better generate output according to the desired output (Saxena [para. 0075]) in a manner that would have yielded predictable results at the relevant time.
wherein the second set of arguments includes at least output data from the first function call or API call. Kim, which is in the same field of endeavor of order fulfillment and commercial interactions, discloses this limitation. (… method may comprise: receiving a return application programming interface (API) call from a user device requesting a return of a returned item; validating the return API call against data records of a networked database; determining that the returned item comprises a plurality of individual items; analyzing a return code of the return API call to determine whether a subset of the individual items is defective; determining a portion of a price of the returned item corresponding to the subset of the individual items; generating a refund API call to issue the portion of the price of the returned item as a refund; updating the data records of the networked database to record the refund; and transmitting a notification to the user device regarding an approval of the return API call. Kim [para. 0007-0008]. … external front end system 103 may run custom web server software designed to receive and process requests from external devices (e.g., mobile device 102A or computer 102B), acquire information from databases and other data stores based on those requests, and provide responses to the received requests based on acquired information. Kim [para. 0025-0028]. … A picker may receive an instruction to place (or “stow”) the items in particular spots in picking zone 209, such as a particular space on a storage unit 210. For example, a picker may scan item 202A using a mobile device (e.g., device 119B). The device may indicate where the picker should stow item 202A, for example, using a system that indicate an aisle, shelf, and location. The device may then prompt the picker to scan a barcode at that location before stowing item 202A in that location. The device may send (e.g., via a wireless network) data to a computer system such as WMS 119 in FIG. 1A indicating that item 202A has been stowed at the location by the user using device 119B. Kim [para. 0060-0066]. … Intake sub-system 300 may be designated for initial processing of a communication from a source application program interface (API) 302. Source API 302 may be any one of a number of APIs, which may be specifically configured for use by a consumer, a delivery-person, an administrator, and/or a seller. Source API 302 may be implemented on a computing device. Kim [para. 0069-0070]). It would have been obvious to one of ordinary skill in the art of automated workflow sequencing and communication protocol before the effective filing date of the claimed invention to modify the data transmission and receipt steps of Koller et al. and Saxena combined to include the second set of arguments includes at least output data from the first function call or API call as disclosed by Kim to enable internal users (e.g., employees of an organization that owns, operates, or leases system) to interact with one or more systems (Kim [para. 0038]) in a manner that would have yielded predictable results at the relevant time.
executing the sequence of actions to complete at least a portion of the task; (… the method can include instructing the corresponding substation to dispense an ingredient or execute an action. The method can also include determining each of the build steps have been completed by a user. Koller et al. [para. 0009]);
wherein executing the sequence of actions comprises: invoking the first function call or API call with the first set of arguments identified from the response, and invoking the second function call or API call with the second set of arguments identified from the response. Saxena discloses this limitation. (the communications facility 729 is configured to provide automated responses to customer requests and/or provide recommendations. Saxena [para. 0095]. … Applications 742A-B may be connected to the commerce management engine 736 through an interface 740A-B (e.g., through REST (REpresentational State Transfer) and/or GraphQL APIs) to expose the functionality and/or data available through and within the commerce management engine 736 to the functionality of applications. … applications 742A-B may utilize APIs to pull data on demand (e.g., customer creation events, product change events, or order cancelation events, etc.) or have the data pushed when updates occur. Saxena [para. 0102-0105]). It would have been obvious to one of ordinary skill in the art of data analysis via generative AI and automated workflow sequencing and communication protocols before the effective filing date of the claimed invention to modify the data analysis steps of Koller et al. and Kim combined to include executing the sequence of actions comprises: invoking the first function call or API call with the first set of arguments identified from the response, and invoking the second function call or API call with the second set of arguments identified from the response as disclosed by Saxena to enable the LLM to better generate output according to the desired output (Saxena [para. 0075]) in a manner that would have yielded predictable results at the relevant time.
and providing results of the execution to the client device. (… the controller can transmit data identifying the multiple drink orders to a computing device and cause display of the drink orders via a display of the computing device based on the data identifying the multiple drink orders. The queue can be displayed on a main display of the beverage station 50. As discussed herein, the beverage preparation station 50 can include a main display or dashboard. The main display or dashboard can also display the progress or status of the multiple drink orders. The main display or dashboard can also display the multiple drink orders in the queue that have been received at block 602 and that are ready to be initiated. Koller et al. [para. 0119]).
Regarding Claim 2, Koller et al., Saxena, and Kim combined disclose the method, wherein the API call follows a REST communication protocol or a gRPC communication protocol. Saxena discloses this limitation. (Applications 742A-B may be connected to the commerce management engine 736 through an interface 740A-B (e.g., through REST (REpresentational State Transfer) and/or GraphQL APIs) to expose the functionality and/or data available through and within the commerce management engine 736 to the functionality of applications. … applications 742A-B may utilize APIs to pull data on demand (e.g., customer creation events, product change events, or order cancelation events, etc.) or have the data pushed when updates occur. Saxena [para. 0102-0105]). It would have been obvious to one of ordinary skill in the art of data analysis via generative AI and automated workflow sequencing and communication protocols before the effective filing date of the claimed invention to modify the data analysis steps of Koller et al. and Kim combined to include the API call follows a REST communication protocol or a gRPC communication protocol as disclosed by Saxena to enable the LLM to better generate output according to the desired output (Saxena [para. 0075]) in a manner that would have yielded predictable results at the relevant time.
Regarding Claim 3, Koller et al., Saxena, and Kim combined disclose the method, wherein the action in the sequence of actions for the workflow follows a schema for the function call or the API call, and wherein an output of the action is an input to a second action in the sequence. (… one or more features of the systems, methods, and devices described herein can utilize a URL and/or cookies, for example for storing and/or transmitting data or user information. … The URLs can include a sequence of characters that identify a path, domain name, a file extension, a host name, a query, a fragment, scheme, a protocol identifier, a port number, a username, a password, a flag, an object, a resource name and/or the like. The systems disclosed herein can generate, receive, transmit, apply, parse, serialize, render, and/or perform an action on a URL. Koller et al. [para. 0144]).
Regarding Claim 4, Koller et al., Saxena, and Kim combined disclose the method, wherein the plan includes at least one checkpoint, and the method further comprising: obtaining intermediate results of executing the plan at the checkpoint; evaluating the intermediate results to identify whether the intermediate results match an expected output; and responsive to determining that the intermediate results match the expected output, executing a remaining portion of the plan. (The beverage preparation station can also track and display steps as they are completed, which can allow the system to predict a time the one or more beverages are to be completed. Koller et al. [para. 0044]. … The beverage production system can also determine if each build step requires a predecessor prior to displaying or allowing execution of the build step, thus preventing errors by the user. Each build step can be tracked and recorded to allow the system to keep track of the steps to build the one or more drink orders correctly and efficiently. Koller et al. [para. 0047]).
Regarding Claim 5, Koller et al., Saxena, and Kim combined disclose the method, wherein the plan includes at least one checkpoint, and the method further comprising: obtaining intermediate results of executing the plan at the checkpoint; evaluating the intermediate results to determine whether the intermediate results match an expected output; (The beverage preparation station can also track and display steps as they are completed, which can allow the system to predict a time the one or more beverages are to be completed. Koller et al. [para. 0044]. … The beverage production system can also determine if each build step requires a predecessor prior to displaying or allowing execution of the build step, thus preventing errors by the user. Each build step can be tracked and recorded to allow the system to keep track of the steps to build the one or more drink orders correctly and efficiently. Koller et al. [para. 0047]).
responsive to determining that the intermediate results do not match the expected output, terminating execution of the plan; and updating the plan to address the intermediate results. ( The portion of the file may include computer-executable instructions that, when executed by a subsystem, cause display of the command at a display at the corresponding substation, dispensing of an ingredient in accordance with the beverage order recipe, and/or waiting for a user to take an action which the corresponding substation can sense. Koller et al. [para. 0067, 0084 (user update status of a build step)]. … the controller may modify the execution of the build steps by delaying execution of particular build steps. Based on the responses to the prompts received from the subsystems, the controller may schedule the execution of the build steps for different times. Koller et al. [para. 0078]. … the substation can determine a build step cannot be completed due to an error. For example, a substation can be out of a certain ingredient, a substation can have a clog along a dispensing line, a substation can have a power issue, a substation can be missing a container (such as a dosing vessel). If a substation determines a build step cannot be completed due to an error, a dashboard (such as the main dashboard or a dashboard of the particular substation) can display an error has occurred and the substation can be prevented from further executing any related steps for the particular build step. Koller et al. [para. 0116]).
Regarding Claim 6, Koller et al., Saxena, and Kim combined disclose the method, wherein the plan includes a second workflow including another sequence of actions. (In response to receiving the prompt, a subsystem can determine whether the subsystem (e.g., all or a portion of the resources of a subsystem) is associated (e.g., currently occupied, scheduled to be occupied, received a prompt associated, etc.) with another build step. All or a portion of the subsystems may be associated with a data store (e.g., a queue, a cache, etc.) identifying commands (or the corresponding build steps for execution). The subsystem can examine the data store and determine if the subsystem is associated with another build step. Koller et al. [para. 0073-0076, 0120]).
Regarding Claim 7, Koller et al., Saxena, and Kim combined disclose the method, wherein the plan includes a checkpoint, and the further comprising: after executing the sequence of actions, receiving one or more user inputs from a user; and executing the another sequence of actions of the second workflow. (… the main display can display the recipes and any customizations for the selected beverage orders, including any or all steps to build the beverage order such as ingredient portions and parameters. In some examples, the main display can receive input from a user to modify the drink orders or the queue for the drink orders. Koller et al. [para. 0091-0092]. … The main display or dashboard can act as a user interface, allowing a user to select multiple beverage orders to start working on simultaneously by selecting it on the queue. … the main display can display the recipes and any customizations for the selected multiple beverage orders, including any or all steps to build the beverage order such as ingredient portions and parameters. In some examples, the main display can receive input from a user to modify the drink orders or the queue for the multiple drink orders. Koller et al. [para. 0120]).
Regarding Claim 9, Koller et al., Saxena, and Kim combined disclose the method, wherein the attributes related to a candidate action in the prompt includes one or a combination of a name of the candidate action, a description of the candidate action, inputs for the candidate action, and expected output type for the candidate action. (The main display or dashboard 52 can act as a user interface, allowing a select one or more beverage orders to start working on by selecting it on the queue. In some examples, the main display or dashboard 52 can be touchscreen or can have one or more buttons. The user can select several beverage orders to work on in parallel as well as in series. … The main display or dashboard 52 can allow a user to modify the drink orders or the queue for the drink orders. Koller et al. [para. 0060]. … he controller can parse data associated with the drink orders, one or more users for building the drink orders, one or more machines for building the drink orders, one or more customers associated with the drink orders, a sequence of steps for building the drink orders, a technique for building the drink orders, one or more ingredients for building the drink orders, or one or more build steps. For example, the data associated with the drink orders may include customer data identifying a customer, step sequence data identifying a sequence of steps, step technique data identifying a technique, ingredient data identifying one or more ingredients, and step data identifying one or more build steps. Koller et al. [para. 0065]).
Regarding Claims 11-17, claims 11-17 recites substantially similar limitations to those of claims 1-7 respectively and are therefore rejected based upon the same prior art combination, reasoning, and rationale. Claims 11-17 recites a non-transitory computer-readable memory storing executable computer program instructions, the instructions executable to perform operations, which is disclosed by Koller et al. [para. 0134]: The elements of a method, process, routine, or algorithm described in connection with the embodiments disclosed herein can be embodied directly in hardware, in a software module executed by a processor device, or in a combination of the two. A software module can reside in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, hard disk, a removable disk, a CD-ROM, or any other form of a non-transitory computer-readable storage medium.
Regarding Claim 20, claim 20 recites substantially similar limitations to those of claim 1 and is therefore rejected based upon the same prior art combination, reasoning, and rationale. Claim 20 recites a computer system, comprising: a computer processor for executing computer program instructions; and a non-transitory computer-readable memory storing executable computer program instructions, the instructions executable to perform operations, which is disclosed by Koller et al. [para. 0014]: a system to prepare one or more beverages can include one or more processors and a computer-readable storage medium including machine-readable instructions that, when executed by the one or more processors, cause the one or more processors to… dynamically generate computer-executable build instructions.
Claims 8 and 18 are rejected under 35 U.S.C. 103 as being unpatentable over Koller et al. (US 2023/0380623), in view of Saxena (US 2024/0330579), in view of Kim et al. (US 2022/0222667), and in further view of Stanley et al. (US 2019/0236525).
Regarding Claim 8, Koller et al., Saxena, and Kim combined disclose the method, further comprising: obtaining a set of data instances, a data instance including a set of inputs describing a task candidate and expected outputs including a desired sequence of actions for completing at least a portion of the task candidate, the desired sequence of actions following a schema for a protocol, (… the machine learning model is configured to generate the sequence based on at least one of customer data or store data. … The system may include and/or utilize artificial intelligence to build the one or more sequences. … Based on the provided input, the machine learning model may output one or more sequences. Further, the machine learning model may be trained using a training data set that includes one or more sequences selected by one or more users and may output the one or more sequences based on the training of the machine learning model. Koller et al. [para. 0016, 0034-0040]. … The beverage production system can also determine if each build step requires a predecessor prior to displaying or allowing execution of the build step, thus preventing errors by the user. Each build step can be tracked and recorded to allow the system to keep track of the steps to build the one or more drink orders correctly and efficiently. Koller et al. [para. 0047]… The machine learning model may be trained to output build steps that are customized (e.g., customized build steps, a customized order of build steps, etc.) for a particular user (e.g., different build steps may be provided for an experienced user v. an inexperienced user). Koller et al. [para. 0064-0065]).
Koller et al., Saxena, and Kim combined fail to explicitly disclose the method generating an estimated output by inputting the set of inputs to the machine-learned model; obtaining a loss function indicating a difference between the estimated output and the expected outputs; and updating parameters of the machine-learned model based on terms obtained from the loss function. Stanley et al. discloses this limitation. (… process of updating a model for predicting the next item to be picked, according to one embodiment. The PMA 112 provides 1110 data identifying the predicted next item to the picker. If the online concierge system 102 predicts the next item … the online concierge system 102 compares 1130 the data identifying the next item picked to the predicted next item to determine whether the picker picked the predicted next item. If the picker did pick the predicted next item, the online concierge system 102 may provide data identifying the predicted next item to the picker … If the picker picked a different item from the predicted next item, the online concierge system 102 (e.g., the modeling engine 220) adjusts 1040 the hidden layers (e.g., the neural network hidden layers 708) and one or more embedding layers (e.g., one or more of the embeddings 218) to improve the model and/or the embeddings to better reflect current picker behavior or layouts. In some embodiments, the online concierge system 102 inputs the predicted next item and picked next item into a cross entropy loss function, which generates any needed updates for the picking sequence model 216 and/or embeddings 218. Stanley et al. [para. 0042-0045, 0073-0076]). It would have been obvious to one of ordinary skill in the art of business modeling and risk management before the effective filing date of the claimed invention to modify the data processing steps of Koller et al., Saxena, and Kim combined to include generating an estimated output by inputting the set of inputs to the machine-learned model; obtaining a loss function indicating a difference between the estimated output and the expected outputs; and updating parameters of the machine-learned model based on terms obtained from the loss function as disclosed by Stanley et al. to periodically transmit an updated model and embeddings (Stanley et al. [para. 0077]), in a manner that would yield predictable results to at the relevant time.
Regarding Claim 18, claim 18 recites substantially similar limitations to those of claim 8 and is therefore rejected based upon the same prior art combination, reasoning, and rationale. Claim 18 recites a non-transitory computer-readable memory storing executable computer program instructions, the instructions executable to perform operations, which is disclosed by Koller et al. [para. 0134]: The elements of a method, process, routine, or algorithm described in connection with the embodiments disclosed herein can be embodied directly in hardware, in a software module executed by a processor device, or in a combination of the two. A software module can reside in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, hard disk, a removable disk, a CD-ROM, or any other form of a non-transitory computer-readable storage medium.
Claim 10 is rejected under 35 U.S.C. 103 as being unpatentable over Koller et al. (US 2023/0380623), in view of Saxena (US 2024/0330579), in view of Kim et al. (US 2022/0222667), and in further view of Volkovs et al. (US 2022/0058489).
Regarding Claim 10, Koller et al., Saxena, and Kim combined fail to explicitly disclose the method, wherein the sequence of actions are represented as a set of nodes. Volkovs et al. discloses this limitation. (… a feedforward neural network may be a multilayer perceptron (MLP) with at least one hidden layer of nodes, where each node may be associated with a weight that is trained and optimized during a training process. During the training process, the weights (or parameters) are optimized through a backpropagation process that aims to minimize a reconstruction error by adjusting (e.g. training) the parameters. The preference decoder 251 may reconstruct preference matrix by generating likelihood scores, which may be used to make predictions such as generating a list of recommended items for the user. Volkovs et al. [para. 0034-0038]). It would have been obvious to one of ordinary skill in the art of business modeling and data processing before the effective filing date of the claimed invention to modify the data processing steps of Koller et al., Saxena, and Kim combined to include the sequence of actions are represented as a set of nodes, as disclosed by Volkovs et al. to increase frequency and quality of interactions between users and the online system (Volkovs et al. [para. 0018]), in a manner that would have yielded predictable results at the relevant time.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure:
Rafferty et al. (US 2022/0004954) – workflow generation system may receive communication data identifying a communication created by a user of a client device, and may process the communication data, with a machine learning model, to determine whether a workflow is needed and particular recipients to be included in the workflow. The machine learning model may be trained based on historical communication data, historical workflow data based on natural language processing, and historical response data based on a sentiment analysis. The workflow generation system may generate a proposed workflow when the workflow is determined to be needed and based on the particular recipients, and may provide data identifying the proposed workflow to the client device.
Tremblay et al. (US 2022/0343250) – methods, systems, components, processes, modules, blocks, circuits, sub-systems, articles, and other elements (collectively referred to in some cases as the “platform” or the “system”) that collectively enable, in one or more datastores and systems. A system and method (e.g., a custom workflow actions system and related processes) for creating and using a custom action (e.g., custom code action). The custom code action may be created and added to a workflow. The workflow may be executed based on the occurrence of one or more events such that the new custom code action may be triggered as part of the workflow (e.g., resulting in an occurrence of a customized action corresponding with a custom instruction code).
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 LETORIA G KNIGHT whose telephone number is (571)270-0485. The examiner can normally be reached M-F 9am-5pm.
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.
/L.G.K/Examiner, Art Unit 3623 /RUTAO WU/Supervisory Patent Examiner, Art Unit 3623
Prosecution Insights