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 .
Claims 1-20 are pending in this application.
Response to Amendment
This Office Action is in response to applicant’s communication filed on January 6th, 2026. The applicant’s remark and amendments to the claims were considered with the results that follow.
In response to the last Office Action, claims 1, 10, and 19 have been amended. No claims have been canceled or added. As a result, claims 1-20 are pending in this application.
Applicant’s argument filed on January 6th, 2026, with respects to claims 1-20 being rejected under 35 U.S.C 101 as being non-statutory subject matter because the claim(s) as a whole are not significantly more than the abstract idea have overcome the rejection. The rejection has been withdrawn due to the rejection filed on January 6th, 2026.
Response to Arguments
Applicant’s arguments, see pg. 16-17, filed on January 6th, 2026, with respect to rejections of claims 1, 10, and 19 under 35 U.S.C 103 have been fully considered but are moot because the new ground of rejection necessitated by applicant’s amendment.
Claim Rejections - 35 USC § 103
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 1-3, 7, 9-12, 16, 18, and 19 are rejected under 35 U.S.C. 103 as being unpatentable over U.S Patent Application Publication 2020/0395007 issued to Cheng et al. (hereinafter as "Cheng") in view of U.S Patent 11,971,914 issued to Watson et al. (hereinafter as “Watson”) in further view of U.S Patent Application Publication 2025/0139263 issued to Sne et al. (hereinafter as “Sne”).
Regarding claim 1, Cheng teaches a method for prompting a generative artificial intelligence (AI) model comprising: generating a query-response message pair comprising a query message comprising an action invocation and a response message comprising information responsive to the action invocation (Cheng: [0120]; The NLU module 904 digests the end user's utterance 902, parses it, and outputs a machine-readable interpretation to the info agent 908. In this example, the understand module 910 cannot understand the end user utterances directly to identify or extract the right information to fulfill the information value directly. [0122]; the survey module 926 conducts a dialog question 928 via a natural language generation (NLG) module to acquire the information value directly from the end user. This usually is the last resort after the understand and infer modules 910 and 916 cannot ascertain the information value.
PNG
media_image1.png
321
537
media_image1.png
Greyscale
{Examiner correlates the query message pair based on the user requesting information by asking the AI model to recommend some jeans on sale in which the AI model will respond back immediately by providing a response accordingly});
obtaining one or more interaction messages associated with interactions between two or more of a user, an assistant service, and a processing system (Cheng: [0108]; the ability to combine the current parsed user input 606 with previously known contextual data from the contextual data store 618 or from other info agents 620 to compute the information value. This could be based on conversation history between the end user and the virtual assistant platform or some user profile data. [0131]; The contextual data store 1004 stores information on “memory,” comprising past interaction history, such as previous dialog turns, conversations from last week, shopper profiles (order history, stated shopping preference, etc.), and so on, and “domain knowledge” comprising the aggregate information that is specific to the services supported by the platform, such as the merchant's product catalog {Examiner correlates the obtaining the entire interactions based on the conversation between the user and assistant service by obtaining from the contextual data store previous conversation history and combining with the current setting to assist the user});
generating a second system message comprising an instruction for the generative AI model to generate an utterance to be provided to the user and an indication of one or more actions available to the generative AI model (Cheng: [0188]; Table 6 shows a dialog example rendered by a single service with multiple action agent(s) connected via a follow-up. Note both the diagram 1600 in FIG. 16 and the dialog sample in Table 6 illustrate the same conversation sequence.
PNG
media_image2.png
505
540
media_image2.png
Greyscale
{Examiner correlates generating a utterance to be provided by the user based on reviewing the user request and the responding accordingly and then give the user additional action to which the user can proceed to follow up after the AI receives additional information});
transmitting, to the generative AI model, a prompt comprising the first system message, the query-response message pair, the one or more interaction messages (Cheng: [0107]-[0108]; The understand module 610 comprises the ability to understand end user utterances in natural language form 606 and identify and extract the right information to fulfill the information value…the info agent 608 is done and can output the information value of interest at step 614. For example, for “Email Info Agent”, the agent will understand the following utterance: “Hi my email address is tom@gmail.com” and automatically fulfill the information value to be “tom@gmail.com”…the ability to combine the current parsed user input 606 with previously known contextual data from the contextual data store 618 or from other info agents 620 to compute the information value. This could be based on conversation history between the end user and the virtual assistant platform or some user profile data), and the second system message (Cheng: [0188]; Table 6 shows a dialog example rendered by a single service with multiple action agent(s) connected via a follow-up. Note both the diagram 1600 in FIG. 16 and the dialog sample in Table 6 illustrate the same conversation sequence.
PNG
media_image2.png
505
540
media_image2.png
Greyscale
{Examiner correlates transmitting the prompt, query response message, and interactions as based on the dialog turn sequence provided in the history in which indicates the transmitting of the set of multi messages to the ai model and the second message based on the indication of available actions according to the utterance requested in which the bot responds back with actions}); and
receiving, from the generative AI model and based at least in part on the prompt, an output of the generative AI model indicating the utterance to be provided to the user by the assistant service (Cheng: [0158]; Table 2 shows a sample dialogue containing a follow-up action agent. In this example, the follow-up action agent (“Membership Status Check Action Agent”) is an instance of the same agent the original triggering action agent (“Membership Status Check Action Agent”).
PNG
media_image3.png
331
556
media_image3.png
Greyscale
{Examiner correlates receiving from the AI model an output of the generative AI model to indicate a request to respond accordingly based on the user request to check a status in which the AI model follows up upon}).
Although, Cheng teaches generating a first system message of a role that is performed by the generative AI model (Cheng: [0107]; The understand module 610 comprises the ability to understand end user utterances in natural language form 606 and identify and extract the right information to fulfill the information value… For example, for “Email Info Agent”, the agent will understand the following utterance: “Hi my email address is tom@gmail.com” and automatically fulfill the information value to be “tom@gmail.com”. [0109]; For example, Email Info Agent can engage with the end user with the following simple dialog sequence: “What is the email you like to use?”, and the end user responds with “My email is tom@gmail.com”. After this exchange, Email Info Agent will determine that the information value is tom@gmail.com {Examiner correlates that the ai model of the info agent will under the following utterance of the user prompt of the user inputting the email address in which the info agent is understanding its role to be an assistant to help the user locate the user email address and communicating with the visitor});
Cheng does not explicitly teach generating a first system message indicative of a role to be performed by the generative AI model.
However, Watson teaches generating a first system message indicative of a role to be performed by the generative AI model (Watson: Col 7, lines 12-16; Generating a prompt may include assigning a role or identity to the model, assigning a task to the model, and placing conditions on the task related to how the model is to generate the comprehensive answer. The role or identity provides the chat bot model with a reference from which to address the query, and a purpose for providing the comprehensive answer).
It would have been obvious to a person of ordinary skill in the art, before the effective filing date of the invention, to modify Cheng (teaches generating a query-response message pair comprising a query message comprising an action invocation and a response message comprising information responsive to the action invocation; obtaining one or more interaction messages associated with interactions between two or more of a user, an assistant service, and a processing system…transmitting, to the generative AI model, a prompt comprising the first system message, the query-response message pair, the one or more interaction messages, and the second system message; and receiving, from the generative AI model and based at least in part on the prompt, an output of the generative AI model indicating the utterance to be provided to the user by the assistant service) with the teachings of Watson (teaches generating a first system message indicative of a role to be performed by the generative AI model). One of ordinary skill in the art would have been motivated to make such a combination of evaluating the user input and improving the accuracy of the output by assigning the bot to a specific topic to address the query accordingly based on the role instructed (See Watson: Col 11, lines 21-24). In addition, the references (Cheng and Watson) teach features that are directed to analogous art and they are directed to the same field of endeavor as Cheng and Watson are directed to natural language conversation system that utilizes assist users accordingly based on the customer requirement.
The modification of Cheng and Watson teaches claimed invention substantially as claimed, however the modification of Cheng and Watson does not explicitly teach dynamically assembling a prompt comprising the first system message, the query-response message pair, the one or more interaction messages, and the second system message, wherein the first system message and the second system message are positioned to receive greater processing weight from the generative AI model than intermediate messages in the prompt;
However, Sne teaches dynamically assembling a prompt comprising the first system message, the query-response message pair, the one or more interaction messages, and the second system message (Sne: [0021]; The techniques may be triggered by a user request for access, by a request by another user, or by automated systems or processes. [0024]; The audit log requests could also be reviewed and labelled to be used as further training data. [0045]; When the LLM is called details of previous decisions made by humans may also be included in the prompts to the LLM (a technique often known as Retrieval Augmented Generation). Those previous decisions may be obtained from the audit log, as explained elsewhere, based on the user, the file requested, or any other parameter which may be related to previous requests. [0064]; The amount of detail stored in the log may be configurable to only store high-level details of the request and outcome, or the specific inputs, outputs and reasons may be stored. [0069]; The training requests situations are captured in documents describing the request, expected outcomes and justifications.
PNG
media_image4.png
502
572
media_image4.png
Greyscale
PNG
media_image5.png
105
452
media_image5.png
Greyscale
{Examiner correlates the User data of the project manager as the first system message, the query response pair as past request and decisions, the one or more interaction message as user data and peers data and file data which the user request, and the second system message as the output as requested from the audit log. The user only provides justification which is the request while the rest of the audit log based on the decision afterwards is dynamically assembled by the LLM as it calls the previous decisions that are included in the prompt based on the request. Sne indicates on [0021], “The techniques may be triggered by a user request for access, by a request by another user, or by automated systems or processes”. Then the LLM automatically process that request which then dynamically assembles the examples as explain on [0045], “When the LLM is called details of previous decisions made by humans may also be included in the prompts to the LLM (a technique often known as Retrieval Augmented Generation)… Those previous decisions may be obtained from the audit log, as explained elsewhere, based on the user, the file requested, or any other parameter which may be related to previous requests”), wherein
the first system message and the second system message are positioned to receive greater processing weight from the generative AI model than intermediate messages in the prompt (Sne: [0021]; The techniques may be triggered by a user request for access, by a request by another user, or by automated systems or processes. [0030]; The weight given to each parameter may be configurable. [0036]; In step 22 a language analysis system, for example a Large Language Model (LLM), is utilized to analyze the content of the user's request and the content of the file to which access is requested….prompt engineering process to provide the required context and related information to the LLM to take a decision. The annotations may include details of which parameters were considered important in the decision, and/or which elements of the natural language request were considered important. [0045]; When the LLM is called details of previous decisions made by humans may also be included in the prompts to the LLM (a technique often known as Retrieval Augmented Generation). Those previous decisions may be obtained from the audit log, as explained elsewhere, based on the user, the file requested, or any other parameter which may be related to previous requests. [0069]; The training requests situations are captured in documents describing the request, expected outcomes and justifications.
PNG
media_image6.png
260
557
media_image6.png
Greyscale
PNG
media_image7.png
153
440
media_image7.png
Greyscale
{Examiner correlates the first system message and the second system message (User data of the role of the project manager and the Output results as high priority based on the high weight given to the user role and past behavior) are positioned to receive greater processing weight due being configured to have high weight while the rest of the message have no high weight based on the user request in the prompt);
It would have been obvious to a person of ordinary skill in the art, before the effective filing date of the invention, to modify Cheng (teaches generating a query-response message pair comprising a query message comprising an action invocation and a response message comprising information responsive to the action invocation; obtaining one or more interaction messages associated with interactions between two or more of a user, an assistant service, and a processing system…transmitting, to the generative AI model, a prompt comprising the first system message, the query-response message pair, the one or more interaction messages, and the second system message; and receiving, from the generative AI model and based at least in part on the prompt, an output of the generative AI model indicating the utterance to be provided to the user by the assistant service) with the teachings of Watson (teaches generating a first system message indicative of a role to be performed by the generative AI model) with the further teachings of Sne (teaches dynamically assembling a prompt comprising the first system message, the query-response message pair, the one or more interaction messages, and the second system message, wherein the first system message and the second system message are positioned to receive greater processing weight from the generative AI model than intermediate messages in the prompt). One of ordinary skill in the art would have been motivated to make such a combination of evaluating the user input and improving the accuracy of the output by improving the efficiency and allowing a more comprehensive and accurate consideration to take place (See Sne: [0031]). In addition, the references (Cheng, Watson, and Sne) teach features that are directed to analogous art and they are directed to the same field of endeavor as Cheng, Watson, and Sne are directed to natural language conversation system that utilizes assist users accordingly based on the customer requirement.
Regarding claim 2, the modification of Cheng, Watson, and Sne teach the claimed invention substantially as claimed, and Cheng further teaches the one or more actions available to the generative AI model comprise retrieval of information from the processing system (Cheng: [0186]; A more complicated example involves a “ Return Action Agent ” that requires three info agents before it can perform any return verification or return slip generation . Note that in practice , a majority of the services fall into this " Single Action with Info Agent ( s ) as Prerequisites " design pattern . Table 5 shows a sample conversation comprising
PNG
media_image8.png
445
552
media_image8.png
Greyscale
);
the second system message comprises an instruction for the generative AI model to determine whether information is to be retrieved from the processing system (Cheng: [0188]; Table 6 shows a dialog example rendered by a single service with multiple action agent(s) connected via a follow-up. Note both the diagram 1600 in FIG. 16 and the dialog sample in Table 6 illustrate the same conversation sequence.
PNG
media_image2.png
505
540
media_image2.png
Greyscale
{Examiner correlates as indicated in table 6 shows if the information is retrievable according to following up to check the status of the account and then giving available actions upon based on the prompt received}); and
the method further comprises receiving a request from the generative AI model to retrieve information from the processing system, and transmitting the retrieved information to the generative AI model (Cheng: [0110]; In summary, the info agent 608 possesses the capability of contextual data retrieval, and computation and automatic sub-dialog generation to engage with end users. It is NLU-enabled, contextual, and conversational. The end-result is the availability of the information value of interest), wherein
the utterance to be provided to the generative AI model is based at least in part on the retrieved information (Cheng: [0110]; In summary, the info agent 608 possesses the capability of contextual data retrieval, and computation and automatic sub-dialog generation to engage with end users. It is NLU-enabled, contextual, and conversational. The end-result is the availability of the information value of interest. [0114]; The understand module 710 understands the end user utterances in natural language form 706 and identifies and extracts the right information to fulfill the information value. That is, if the understand module 710 ascertains the information value from the utterance in decision step 712, then the info agent 708 is done and can output the information value of interest at step 714, as output information value 730), and wherein
the one or more interaction messages comprise the request from the generative AI model to retrieve the information from the processing system and an indication of the transmission of the retrieved information to the generative AI model (Cheng: [0235]; Table 8 shows a multi - service sample dialog from the masterbot's point of view corresponding to the sample multi - service conversation illustrated in diagram 2000.
PNG
media_image9.png
555
582
media_image9.png
Greyscale
).
Regarding claim 3, the modification of Cheng, Watson, and Sne teach the claimed invention substantially as claimed, and Cheng further teaches the one or more actions available to the generative AI model comprise prompting the user for additional information (Cheng: [0186]; A more complicated example involves a “ Return Action Agent ” that requires three info agents before it can perform any return verification or return slip generation . Note that in practice , a majority of the services fall into this " Single Action with Info Agent ( s ) as Prerequisites " design pattern . Table 5 shows a sample conversation comprising
PNG
media_image8.png
445
552
media_image8.png
Greyscale
);
the second system message comprises an instruction for the generative AI model to determine whether the additional information is to be requested from the user (Cheng: [0186]; A more complicated example involves a “ Return Action Agent ” that requires three info agents before it can perform any return verification or return slip generation . Note that in practice , a majority of the services fall into this " Single Action with Info Agent ( s ) as Prerequisites " design pattern . Table 5 shows a sample conversation comprising
PNG
media_image8.png
445
552
media_image8.png
Greyscale
); and
the method further comprises receiving a request from the generative AI model to request the additional information from the user, and transmitting the additional information to the generative AI model (Cheng: [0186]; A more complicated example involves a “ Return Action Agent ” that requires three info agents before it can perform any return verification or return slip generation . Note that in practice , a majority of the services fall into this " Single Action with Info Agent ( s ) as Prerequisites " design pattern . Table 5 shows a sample conversation comprising
PNG
media_image8.png
445
552
media_image8.png
Greyscale
), wherein
the utterance to be provided to the generative AI model is based at least in part on the additional information (Cheng: [0186]; A more complicated example involves a “ Return Action Agent ” that requires three info agents before it can perform any return verification or return slip generation . Note that in practice , a majority of the services fall into this " Single Action with Info Agent ( s ) as Prerequisites " design pattern . Table 5 shows a sample conversation comprising
PNG
media_image8.png
445
552
media_image8.png
Greyscale
), and
wherein the one or more interaction messages comprise the request from the generative AI model to request the additional information from the user and an indication of the transmission of the additional information to the generative AI model (Cheng: [0186]; A more complicated example involves a “ Return Action Agent ” that requires three info agents before it can perform any return verification or return slip generation . Note that in practice , a majority of the services fall into this " Single Action with Info Agent ( s ) as Prerequisites " design pattern . Table 5 shows a sample conversation comprising
PNG
media_image8.png
445
552
media_image8.png
Greyscale
).
Regarding claim 7, the modification of Cheng, Watson, and Sne teach the claimed invention substantially as claimed, and Cheng further teaches the query-response message pair is selected based at least in part on a context in which the generative AI model is to produce the output of the generative AI model (Cheng: [0186]; A more complicated example involves a “ Return Action Agent ” that requires three info agents before it can perform any return verification or return slip generation . Note that in practice , a majority of the services fall into this " Single Action with Info Agent ( s ) as Prerequisites " design pattern . Table 5 shows a sample conversation comprising
PNG
media_image8.png
445
552
media_image8.png
Greyscale
).
Regarding claim 9, the modification of Cheng, Watson, and Sne teach
the claimed invention substantially as claimed, and Cheng further teaches interactions between the user and the assistant service are expressed in plain language, interactions between the generative AI model and the assistant service are expressed in a structured data format, and interactions between the assistant service and the processing system are expressed in the structured data format (Cheng: [0111]; the more new services the platform can quickly assemble to expand the usefulness of the conversation bot. The implementation of an info agent depends on the actual embodiment of the overall virtual assistant platform. For example, an info agent can specify using JSON or XML, what entity extraction type and value it should focus on, and how to map that information to the final value. The XML can also specify how to combine different information sources to make an inference, what other related info agent(s) it can leverage, and how to deduce the information value from those related info agents, and finally, how to compose the survey question to ask the end user for obtaining additional information. [0158]; Table 2 shows a sample dialogue containing a follow-up action agent. In this example, the follow-up action agent (“Membership Status Check Action Agent”) is an instance of the same agent the original triggering action agent (“Membership Status Check Action Agent”).
PNG
media_image3.png
331
556
media_image3.png
Greyscale
).
Regarding claim 10, Cheng teaches an apparatus, comprising: one or more memories storing processor-executable code (Cheng: [0290]; The computer programs typically comprise one or more instructions set at various times in various memory and storage devices in a computing device or computer, and that, when read and executed by one or more processors); and one or more processors coupled with the one or more memories and individually or collectively operable to execute the code to cause the apparatus to (Cheng: [0290]; The computer programs typically comprise one or more instructions set at various times in various memory and storage devices in a computing device or computer, and that, when read and executed by one or more processors)): generate a query-response message pair comprising a query message comprising an action invocation and a response message comprising information responsive to the action invocation (Cheng: [0120]; The NLU module 904 digests the end user's utterance 902, parses it, and outputs a machine-readable interpretation to the info agent 908. In this example, the understand module 910 cannot understand the end user utterances directly to identify or extract the right information to fulfill the information value directly. [0122]; the survey module 926 conducts a dialog question 928 via a natural language generation (NLG) module to acquire the information value directly from the end user. This usually is the last resort after the understand and infer modules 910 and 916 cannot ascertain the information value.
PNG
media_image10.png
236
523
media_image10.png
Greyscale
);
obtain one or more interaction messages associated with interactions between two or more of a user, an assistant service, and a processing system (Cheng: [0108]; the ability to combine the current parsed user input 606 with previously known contextual data from the contextual data store 618 or from other info agents 620 to compute the information value. This could be based on conversation history between the end user and the virtual assistant platform or some user profile data. [0131]; The contextual data store 1004 stores information on “memory,” comprising past interaction history, such as previous dialog turns, conversations from last week, shopper profiles (order history, stated shopping preference, etc.), and so on, and “domain knowledge” comprising the aggregate information that is specific to the services supported by the platform, such as the merchant's product catalog);
generate a second system message comprising an instruction for the generative AI model to generate an utterance to be provided to the user and an indication of one or more actions available to the generative AI model (Cheng: [0188]; Table 6 shows a dialog example rendered by a single service with multiple action agent(s) connected via a follow-up. Note both the diagram 1600 in FIG. 16 and the dialog sample in Table 6 illustrate the same conversation sequence.
PNG
media_image2.png
505
540
media_image2.png
Greyscale
);
transmit, to the generative AI model, a prompt comprising the first system message, the query-response message pair, the one or more interaction messages (Cheng: [0107]-[0108]; The understand module 610 comprises the ability to understand end user utterances in natural language form 606 and identify and extract the right information to fulfill the information value…the info agent 608 is done and can output the information value of interest at step 614. For example, for “Email Info Agent”, the agent will understand the following utterance: “Hi my email address is tom@gmail.com” and automatically fulfill the information value to be “tom@gmail.com”…the ability to combine the current parsed user input 606 with previously known contextual data from the contextual data store 618 or from other info agents 620 to compute the information value. This could be based on conversation history between the end user and the virtual assistant platform or some user profile data), and
the second system message (Cheng: [0188]; Table 6 shows a dialog example rendered by a single service with multiple action agent(s) connected via a follow-up. Note both the diagram 1600 in FIG. 16 and the dialog sample in Table 6 illustrate the same conversation sequence.
PNG
media_image2.png
505
540
media_image2.png
Greyscale
); and
receive, from the generative AI model and based at least in part on the prompt, an output of the generative AI model indicating the utterance to be provided to the user by the assistant service (Cheng: [0158]; Table 2 shows a sample dialogue containing a follow-up action agent. In this example, the follow-up action agent (“Membership Status Check Action Agent”) is an instance of the same agent the original triggering action agent (“Membership Status Check Action Agent”).
PNG
media_image3.png
331
556
media_image3.png
Greyscale
).
Although, Cheng teaches generating a first system message of a role that is performed by the generative AI model (Cheng: [0107]; The understand module 610 comprises the ability to understand end user utterances in natural language form 606 and identify and extract the right information to fulfill the information value… For example, for “Email Info Agent”, the agent will understand the following utterance: “Hi my email address is tom@gmail.com” and automatically fulfill the information value to be “tom@gmail.com”. [0109]; For example, Email Info Agent can engage with the end user with the following simple dialog sequence: “What is the email you like to use?”, and the end user responds with “My email is tom@gmail.com”. After this exchange, Email Info Agent will determine that the information value is tom@gmail.com {Examiner correlates that the ai model of the info agent will under the following utterance of the user prompt of the user inputting the email address in which the info agent is understanding its role to be an assistant to help the user locate the user email address and communicating with the visitor});
Cheng does not explicitly teach generating a first system message indicative of a role to be performed by the generative AI model.
However, Watson teaches generating a first system message indicative of a role to be performed by the generative AI model (Watson: Col 7, lines 12-16; Generating a prompt may include assigning a role or identity to the model, assigning a task to the model, and placing conditions on the task related to how the model is to generate the comprehensive answer. The role or identity provides the chat bot model with a reference from which to address the query, and a purpose for providing the comprehensive answer).
It would have been obvious to a person of ordinary skill in the art, before the effective filing date of the invention, to modify Cheng (teaches generating a query-response message pair comprising a query message comprising an action invocation and a response message comprising information responsive to the action invocation; obtaining one or more interaction messages associated with interactions between two or more of a user, an assistant service, and a processing system…transmitting, to the generative AI model, a prompt comprising the first system message, the query-response message pair, the one or more interaction messages, and the second system message; and receiving, from the generative AI model and based at least in part on the prompt, an output of the generative AI model indicating the utterance to be provided to the user by the assistant service) with the teachings of Watson (teaches generating a first system message indicative of a role to be performed by the generative AI model). One of ordinary skill in the art would have been motivated to make such a combination of evaluating the user input and improving the accuracy of the output by assigning the bot to a specific topic to address the query accordingly based on the role instructed (See Watson: Col 11, lines 21-24). In addition, the references (Cheng and Watson) teach features that are directed to analogous art and they are directed to the same field of endeavor as Cheng and Watson are directed to natural language conversation system that utilizes assist users accordingly based on the customer requirement.
The modification of Cheng and Watson teaches claimed invention substantially as claimed, however the modification of Cheng and Watson does not explicitly teach dynamically assembling a prompt comprising the first system message, the query-response message pair, the one or more interaction messages, and the second system message, wherein the first system message and the second system message are positioned to receive greater processing weight from the generative AI model than intermediate messages in the prompt;
However, Sne teaches dynamically assembling a prompt comprising the first system message, the query-response message pair, the one or more interaction messages, and the second system message (Sne: [0021]; The techniques may be triggered by a user request for access, by a request by another user, or by automated systems or processes. [0024]; The audit log requests could also be reviewed and labelled to be used as further training data. [0045]; When the LLM is called details of previous decisions made by humans may also be included in the prompts to the LLM (a technique often known as Retrieval Augmented Generation). Those previous decisions may be obtained from the audit log, as explained elsewhere, based on the user, the file requested, or any other parameter which may be related to previous requests. [0064]; The amount of detail stored in the log may be configurable to only store high-level details of the request and outcome, or the specific inputs, outputs and reasons may be stored. [0069]; The training requests situations are captured in documents describing the request, expected outcomes and justifications.
PNG
media_image4.png
502
572
media_image4.png
Greyscale
PNG
media_image5.png
105
452
media_image5.png
Greyscale
{Examiner correlates the User data of the project manager as the first system message, the query response pair as past request and decisions, the one or more interaction message as user data and peers data and file data which the user request, and the second system message as the output as requested from the audit log. The user only provides justification which is the request while the rest of the audit log based on the decision afterwards is dynamically assembled by the LLM as it calls the previous decisions that are included in the prompt based on the request. Sne indicates on [0021], “The techniques may be triggered by a user request for access, by a request by another user, or by automated systems or processes”. Then the LLM automatically process that request which then dynamically assembles the examples as explain on [0045], “When the LLM is called details of previous decisions made by humans may also be included in the prompts to the LLM (a technique often known as Retrieval Augmented Generation)… Those previous decisions may be obtained from the audit log, as explained elsewhere, based on the user, the file requested, or any other parameter which may be related to previous requests”), wherein
the first system message and the second system message are positioned to receive greater processing weight from the generative AI model than intermediate messages in the prompt (Sne: [0021]; The techniques may be triggered by a user request for access, by a request by another user, or by automated systems or processes. [0030]; The weight given to each parameter may be configurable. [0036]; In step 22 a language analysis system, for example a Large Language Model (LLM), is utilized to analyze the content of the user's request and the content of the file to which access is requested….prompt engineering process to provide the required context and related information to the LLM to take a decision. The annotations may include details of which parameters were considered important in the decision, and/or which elements of the natural language request were considered important. [0045]; When the LLM is called details of previous decisions made by humans may also be included in the prompts to the LLM (a technique often known as Retrieval Augmented Generation). Those previous decisions may be obtained from the audit log, as explained elsewhere, based on the user, the file requested, or any other parameter which may be related to previous requests. [0069]; The training requests situations are captured in documents describing the request, expected outcomes and justifications.
PNG
media_image6.png
260
557
media_image6.png
Greyscale
PNG
media_image7.png
153
440
media_image7.png
Greyscale
{Examiner correlates the first system message and the second system message (User data of the role of the project manager and the Output results as high priority based on the high weight given to the user role and past behavior) are positioned to receive greater processing weight due being configured to have high weight while the rest of the message have no high weight based on the user request in the prompt);
It would have been obvious to a person of ordinary skill in the art, before the effective filing date of the invention, to modify Cheng (teaches generating a query-response message pair comprising a query message comprising an action invocation and a response message comprising information responsive to the action invocation; obtaining one or more interaction messages associated with interactions between two or more of a user, an assistant service, and a processing system…transmitting, to the generative AI model, a prompt comprising the first system message, the query-response message pair, the one or more interaction messages, and the second system message; and receiving, from the generative AI model and based at least in part on the prompt, an output of the generative AI model indicating the utterance to be provided to the user by the assistant service) with the teachings of Watson (teaches generating a first system message indicative of a role to be performed by the generative AI model) with the further teachings of Sne (teaches dynamically assembling a prompt comprising the first system message, the query-response message pair, the one or more interaction messages, and the second system message, wherein the first system message and the second system message are positioned to receive greater processing weight from the generative AI model than intermediate messages in the prompt). One of ordinary skill in the art would have been motivated to make such a combination of evaluating the user input and improving the accuracy of the output by improving the efficiency and allowing a more comprehensive and accurate consideration to take place (See Sne: [0031]). In addition, the references (Cheng, Watson, and Sne) teach features that are directed to analogous art and they are directed to the same field of endeavor as Cheng, Watson, and Sne are directed to natural language conversation system that utilizes assist users accordingly based on the customer requirement.
Regarding claim 11, the modification of Cheng, Watson, and Sne teach
the claimed invention substantially as claimed, and Cheng further teaches the one or more actions available to the generative AI model comprise retrieval of information from the processing system (Cheng: [0186]; A more complicated example involves a “ Return Action Agent ” that requires three info agents before it can perform any return verification or return slip generation . Note that in practice , a majority of the services fall into this " Single Action with Info Agent ( s ) as Prerequisites " design pattern . Table 5 shows a sample conversation comprising
PNG
media_image8.png
445
552
media_image8.png
Greyscale
);
the second system message comprises an instruction for the generative AI model to determine whether information is to be retrieved from the processing system (Cheng: [0188]; Table 6 shows a dialog example rendered by a single service with multiple action agent(s) connected via a follow-up. Note both the diagram 1600 in FIG. 16 and the dialog sample in Table 6 illustrate the same conversation sequence.
PNG
media_image2.png
505
540
media_image2.png
Greyscale
); and
the one or more processors are individually or collectively further operable to execute the code to cause the apparatus to receive a request from the generative AI model to retrieve information from the processing system, and transmit the retrieved information to the generative AI model (Cheng: [0110]; In summary, the info agent 608 possesses the capability of contextual data retrieval, and computation and automatic sub-dialog generation to engage with end users. It is NLU-enabled, contextual, and conversational. The end-result is the availability of the information value of interest), wherein
the utterance to be provided to the generative AI model is based at least in part on the retrieved information (Cheng: [0110]; In summary, the info agent 608 possesses the capability of contextual data retrieval, and computation and automatic sub-dialog generation to engage with end users. It is NLU-enabled, contextual, and conversational. The end-result is the availability of the information value of interest. [0114]; The understand module 710 understands the end user utterances in natural language form 706 and identifies and extracts the right information to fulfill the information value. That is, if the understand module 710 ascertains the information value from the utterance in decision step 712, then the info agent 708 is done and can output the information value of interest at step 714, as output information value 730), and wherein
the one or more interaction messages comprise the request from the generative AI model to retrieve the information from the processing system and an indication of the transmission of the retrieved information to the generative AI model (Cheng: [0235]; Table 8 shows a multi - service sample dialog from the masterbot's point of view corresponding to the sample multi - service conversation illustrated in diagram 2000.
PNG
media_image9.png
555
582
media_image9.png
Greyscale
).
Regarding claim 12, the modification of Cheng, Watson, and Sne teach
the claimed invention substantially as claimed, and Cheng further teaches the one or more actions available to the generative AI model comprise prompting the user for additional information (Cheng: [0186]; A more complicated example involves a “ Return Action Agent ” that requires three info agents before it can perform any return verification or return slip generation . Note that in practice , a majority of the services fall into this " Single Action with Info Agent ( s ) as Prerequisites " design pattern . Table 5 shows a sample conversation comprising
PNG
media_image8.png
445
552
media_image8.png
Greyscale
);
the second system message comprises an instruction for the generative AI model to determine whether the additional information is to be requested from the user (Cheng: [0186]; A more complicated example involves a “ Return Action Agent ” that requires three info agents before it can perform any return verification or return slip generation . Note that in practice , a majority of the services fall into this " Single Action with Info Agent ( s ) as Prerequisites " design pattern . Table 5 shows a sample conversation comprising
PNG
media_image8.png
445
552
media_image8.png
Greyscale
); and
the one or more processors are individually or collectively further operable to execute the code to cause the apparatus to receive a request from the generative AI model to request the additional information from the user, and
transmit the additional information to the generative AI model (Cheng: [0186]; A more complicated example involves a “ Return Action Agent ” that requires three info agents before it can perform any return verification or return slip generation . Note that in practice , a majority of the services fall into this " Single Action with Info Agent ( s ) as Prerequisites " design pattern . Table 5 shows a sample conversation comprising
PNG
media_image8.png
445
552
media_image8.png
Greyscale
), wherein
the utterance to be provided to the generative AI model is based at least in part on the additional information (Cheng: [0186]; A more complicated example involves a “ Return Action Agent ” that requires three info agents before it can perform any return verification or return slip generation . Note that in practice , a majority of the services fall into this " Single Action with Info Agent ( s ) as Prerequisites " design pattern . Table 5 shows a sample conversation comprising
PNG
media_image8.png
445
552
media_image8.png
Greyscale
), and wherein
the one or more interaction messages comprise the request from the generative AI model to request the additional information from the user and an indication of the transmission of the additional information to the generative AI model (Cheng: [0186]; A more complicated example involves a “ Return Action Agent ” that requires three info agents before it can perform any return verification or return slip generation . Note that in practice , a majority of the services fall into this " Single Action with Info Agent ( s ) as Prerequisites " design pattern . Table 5 shows a sample conversation comprising
PNG
media_image8.png
445
552
media_image8.png
Greyscale
).
Regarding claim 16, the modification of Cheng, Watson, and Sne teach
the claimed invention substantially as claimed, and Cheng further teaches the query-response message pair is selected based at least in part on a context in which the generative AI model is to produce the output of the generative AI model (Cheng: [0186]; A more complicated example involves a “ Return Action Agent ” that requires three info agents before it can perform any return verification or return slip generation . Note that in practice , a majority of the services fall into this " Single Action with Info Agent ( s ) as Prerequisites " design pattern . Table 5 shows a sample conversation comprising
PNG
media_image8.png
445
552
media_image8.png
Greyscale
).
Regarding claim 18, the modification of Cheng, Watson, and Sne teach the claimed invention substantially as claimed, and Cheng further teaches interactions between the user and the assistant service are expressed in plain language, interactions between the generative AI model and the assistant service are expressed in a structured data format, and interactions between the assistant service and the processing system are expressed in the structured data format (Cheng: [0111]; the more new services the platform can quickly assemble to expand the usefulness of the conversation bot. The implementation of an info agent depends on the actual embodiment of the overall virtual assistant platform. For example, an info agent can specify using JSON or XML, what entity extraction type and value it should focus on, and how to map that information to the final value. The XML can also specify how to combine different information sources to make an inference, what other related info agent(s) it can leverage, and how to deduce the information value from those related info agents, and finally, how to compose the survey question to ask the end user for obtaining additional information. [0158]; Table 2 shows a sample dialogue containing a follow-up action agent. In this example, the follow-up action agent (“Membership Status Check Action Agent”) is an instance of the same agent the original triggering action agent (“Membership Status Check Action Agent”).
PNG
media_image3.png
331
556
media_image3.png
Greyscale
).
Regarding claim 19, Cheng teaches a non-transitory computer-readable medium storing code, the code comprising instructions executable by one or more processors to (Cheng: [0258]; The present invention may be implemented using one or more machine learning modules implementing one or more algorithms implemented in non-transitory storage medium having program code stored thereon, the program code executable by one or more processors): generate a query-response message pair comprising a query message comprising an action invocation and a response message comprising information responsive to the action invocation (Cheng: [0120]; The NLU module 904 digests the end user's utterance 902, parses it, and outputs a machine-readable interpretation to the info agent 908. In this example, the understand module 910 cannot understand the end user utterances directly to identify or extract the right information to fulfill the information value directly. [0122]; the survey module 926 conducts a dialog question 928 via a natural language generation (NLG) module to acquire the information value directly from the end user. This usually is the last resort after the understand and infer modules 910 and 916 cannot ascertain the information value.
PNG
media_image10.png
236
523
media_image10.png
Greyscale
);
obtain one or more interaction messages associated with interactions between two or more of a user, an assistant service, and a processing system (Cheng: [0108]; the ability to combine the current parsed user input 606 with previously known contextual data from the contextual data store 618 or from other info agents 620 to compute the information value. This could be based on conversation history between the end user and the virtual assistant platform or some user profile data. [0131]; The contextual data store 1004 stores information on “memory,” comprising past interaction history, such as previous dialog turns, conversations from last week, shopper profiles (order history, stated shopping preference, etc.), and so on, and “domain knowledge” comprising the aggregate information that is specific to the services supported by the platform, such as the merchant's product catalog);
generate a second system message comprising an instruction for the generative AI model to generate an utterance to be provided to the user and an indication of one or more actions available to the generative AI model (Cheng: [0188]; Table 6 shows a dialog example rendered by a single service with multiple action agent(s) connected via a follow-up. Note both the diagram 1600 in FIG. 16 and the dialog sample in Table 6 illustrate the same conversation sequence.
PNG
media_image2.png
505
540
media_image2.png
Greyscale
);
transmit, to the generative AI model, a prompt comprising the first system message, the query-response message pair, the one or more interaction messages (Cheng: [0107]-[0108]; The understand module 610 comprises the ability to understand end user utterances in natural language form 606 and identify and extract the right information to fulfill the information value…the info agent 608 is done and can output the information value of interest at step 614. For example, for “Email Info Agent”, the agent will understand the following utterance: “Hi my email address is tom@gmail.com” and automatically fulfill the information value to be “tom@gmail.com”…the ability to combine the current parsed user input 606 with previously known contextual data from the contextual data store 618 or from other info agents 620 to compute the information value. This could be based on conversation history between the end user and the virtual assistant platform or some user profile data), and
the second system message (Cheng: [0188]; Table 6 shows a dialog example rendered by a single service with multiple action agent(s) connected via a follow-up. Note both the diagram 1600 in FIG. 16 and the dialog sample in Table 6 illustrate the same conversation sequence.
PNG
media_image2.png
505
540
media_image2.png
Greyscale
); and
receive, from the generative AI model and based at least in part on the prompt, an output of the generative AI model indicating the utterance to be provided to the user by the assistant service (Cheng: [0158]; Table 2 shows a sample dialogue containing a follow-up action agent. In this example, the follow-up action agent (“Membership Status Check Action Agent”) is an instance of the same agent the original triggering action agent (“Membership Status Check Action Agent”).
PNG
media_image3.png
331
556
media_image3.png
Greyscale
).
Although, Cheng teaches generating a first system message of a role that is performed by the generative AI model (Cheng: [0107]; The understand module 610 comprises the ability to understand end user utterances in natural language form 606 and identify and extract the right information to fulfill the information value… For example, for “Email Info Agent”, the agent will understand the following utterance: “Hi my email address is tom@gmail.com” and automatically fulfill the information value to be “tom@gmail.com”. [0109]; For example, Email Info Agent can engage with the end user with the following simple dialog sequence: “What is the email you like to use?”, and the end user responds with “My email is tom@gmail.com”. After this exchange, Email Info Agent will determine that the information value is tom@gmail.com {Examiner correlates that the ai model of the info agent will under the following utterance of the user prompt of the user inputting the email address in which the info agent is understanding its role to be an assistant to help the user locate the user email address and communicating with the visitor});
Cheng does not explicitly teach generating a first system message indicative of a role to be performed by the generative AI model.
However, Watson teaches generating a first system message indicative of a role to be performed by the generative AI model (Watson: Col 7, lines 12-16; Generating a prompt may include assigning a role or identity to the model, assigning a task to the model, and placing conditions on the task related to how the model is to generate the comprehensive answer. The role or identity provides the chat bot model with a reference from which to address the query, and a purpose for providing the comprehensive answer).
It would have been obvious to a person of ordinary skill in the art, before the effective filing date of the invention, to modify Cheng (teaches generating a query-response message pair comprising a query message comprising an action invocation and a response message comprising information responsive to the action invocation; obtaining one or more interaction messages associated with interactions between two or more of a user, an assistant service, and a processing system…transmitting, to the generative AI model, a prompt comprising the first system message, the query-response message pair, the one or more interaction messages, and the second system message; and receiving, from the generative AI model and based at least in part on the prompt, an output of the generative AI model indicating the utterance to be provided to the user by the assistant service) with the teachings of Watson (teaches generating a first system message indicative of a role to be performed by the generative AI model). One of ordinary skill in the art would have been motivated to make such a combination of evaluating the user input and improving the accuracy of the output by assigning the bot to a specific topic to address the query accordingly based on the role instructed (See Watson: Col 11, lines 21-24). In addition, the references (Cheng and Watson) teach features that are directed to analogous art and they are directed to the same field of endeavor as Cheng and Watson are directed to natural language conversation system that utilizes assist users accordingly based on the customer requirement.
The modification of Cheng and Watson teaches claimed invention substantially as claimed, however the modification of Cheng and Watson does not explicitly teach dynamically assembling a prompt comprising the first system message, the query-response message pair, the one or more interaction messages, and the second system message, wherein the first system message and the second system message are positioned to receive greater processing weight from the generative AI model than intermediate messages in the prompt;
However, Sne teaches dynamically assembling a prompt comprising the first system message, the query-response message pair, the one or more interaction messages, and the second system message (Sne: [0021]; The techniques may be triggered by a user request for access, by a request by another user, or by automated systems or processes. [0024]; The audit log requests could also be reviewed and labelled to be used as further training data. [0045]; When the LLM is called details of previous decisions made by humans may also be included in the prompts to the LLM (a technique often known as Retrieval Augmented Generation). Those previous decisions may be obtained from the audit log, as explained elsewhere, based on the user, the file requested, or any other parameter which may be related to previous requests. [0064]; The amount of detail stored in the log may be configurable to only store high-level details of the request and outcome, or the specific inputs, outputs and reasons may be stored. [0069]; The training requests situations are captured in documents describing the request, expected outcomes and justifications.
PNG
media_image4.png
502
572
media_image4.png
Greyscale
PNG
media_image5.png
105
452
media_image5.png
Greyscale
{Examiner correlates the User data of the project manager as the first system message, the query response pair as past request and decisions, the one or more interaction message as user data and peers data and file data which the user request, and the second system message as the output as requested from the audit log. The user only provides justification which is the request while the rest of the audit log based on the decision afterwards is dynamically assembled by the LLM as it calls the previous decisions that are included in the prompt based on the request. Sne indicates on [0021], “The techniques may be triggered by a user request for access, by a request by another user, or by automated systems or processes”. Then the LLM automatically process that request which then dynamically assembles the examples as explain on [0045], “When the LLM is called details of previous decisions made by humans may also be included in the prompts to the LLM (a technique often known as Retrieval Augmented Generation)… Those previous decisions may be obtained from the audit log, as explained elsewhere, based on the user, the file requested, or any other parameter which may be related to previous requests”), wherein
the first system message and the second system message are positioned to receive greater processing weight from the generative AI model than intermediate messages in the prompt (Sne: [0021]; The techniques may be triggered by a user request for access, by a request by another user, or by automated systems or processes. [0030]; The weight given to each parameter may be configurable. [0036]; In step 22 a language analysis system, for example a Large Language Model (LLM), is utilized to analyze the content of the user's request and the content of the file to which access is requested….prompt engineering process to provide the required context and related information to the LLM to take a decision. The annotations may include details of which parameters were considered important in the decision, and/or which elements of the natural language request were considered important. [0045]; When the LLM is called details of previous decisions made by humans may also be included in the prompts to the LLM (a technique often known as Retrieval Augmented Generation). Those previous decisions may be obtained from the audit log, as explained elsewhere, based on the user, the file requested, or any other parameter which may be related to previous requests. [0069]; The training requests situations are captured in documents describing the request, expected outcomes and justifications.
PNG
media_image6.png
260
557
media_image6.png
Greyscale
PNG
media_image7.png
153
440
media_image7.png
Greyscale
{Examiner correlates the first system message and the second system message (User data of the role of the project manager and the Output results as high priority based on the high weight given to the user role and past behavior) are positioned to receive greater processing weight due being configured to have high weight while the rest of the message have no high weight based on the user request in the prompt);
It would have been obvious to a person of ordinary skill in the art, before the effective filing date of the invention, to modify Cheng (teaches generating a query-response message pair comprising a query message comprising an action invocation and a response message comprising information responsive to the action invocation; obtaining one or more interaction messages associated with interactions between two or more of a user, an assistant service, and a processing system…transmitting, to the generative AI model, a prompt comprising the first system message, the query-response message pair, the one or more interaction messages, and the second system message; and receiving, from the generative AI model and based at least in part on the prompt, an output of the generative AI model indicating the utterance to be provided to the user by the assistant service) with the teachings of Watson (teaches generating a first system message indicative of a role to be performed by the generative AI model) with the further teachings of Sne (teaches dynamically assembling a prompt comprising the first system message, the query-response message pair, the one or more interaction messages, and the second system message, wherein the first system message and the second system message are positioned to receive greater processing weight from the generative AI model than intermediate messages in the prompt). One of ordinary skill in the art would have been motivated to make such a combination of evaluating the user input and improving the accuracy of the output by improving the efficiency and allowing a more comprehensive and accurate consideration to take place (See Sne: [0031]). In addition, the references (Cheng, Watson, and Sne) teach features that are directed to analogous art and they are directed to the same field of endeavor as Cheng, Watson, and Sne are directed to natural language conversation system that utilizes assist users accordingly based on the customer requirement.
Claims 4 and 13 are rejected under 35 U.S.C. 103 as being unpatentable over U.S Patent Application Publication 2020/0395007 issued to Cheng et al. (hereinafter as "Cheng") in view of U.S Patent 11,971,914 issued to Watson et al. (hereinafter as “Watson”) in view of U.S Patent Application Publication 2025/0139263 issued to Sne et al. (hereinafter as “Sne”) in further view of U.S Patent Application Publication 2024/0320444 issued to MASCHMEYER et al. (hereinafter as “MASCHMEYER”).
Regarding claim 4, the modification of Cheng, Watson, and Sne teach the claimed invention substantially as claimed, however the modification of Cheng, Watson, and Sne does not explicitly teach the one or more actions available to the generative AI model comprise prompting the user for confirmation of an action selected from the one or more actions available to the generative AI model; the second system message comprises an instruction for the generative AI model to determine whether selected actions are flagged for user confirmation; and the method further comprises receiving a request from the generative AI model to request confirmation of the action from the user, and transmitting the confirmation to the generative AI model, wherein the utterance to be provided to the generative AI model is based at least in part on the confirmation, and wherein the one or more interaction messages comprise the request from the generative AI model to request the confirmation from the user and an indication of the transmission of the confirmation to the generative AI model.
Maschmeyer teaches the one or more actions available to the generative AI model comprise prompting the user for confirmation of an action selected from the one or more actions available to the generative AI model (Maschmeyer: [0102]: In particular, the second output may be an output of the generative model that includes representations of user-selected portions from the outputs of preceding iteration(s) of content generation. The second output data may be presented to the user via the user interface. In some implementations, the computing system may prompt the user to confirm the second output data as the final content output or to make further selections of portions from the second output(s));
the second system message comprises an instruction for the generative AI model to determine whether selected actions are flagged for user confirmation (Maschmeyer: [0102]: In particular, the second output may be an output of the generative model that includes representations of user-selected portions from the outputs of preceding iteration(s) of content generation. The second output data may be presented to the user via the user interface. In some implementations, the computing system may prompt the user to confirm the second output data as the final content output or to make further selections of portions from the second output(s). This process of iteratively modifying input prompts to the generative model and producing new content outputs may be continued until the user either flags one of the outputs as the final content output or manually terminates the process); and
the method further comprises receiving a request from the generative AI model to request confirmation of the action from the user, and transmitting the confirmation to the generative AI model, wherein the utterance to be provided to the generative AI model is based at least in part on the confirmation (Maschmeyer: [0102]; This process of iteratively modifying input prompts to the generative model and producing new content outputs may be continued until the user either flags one of the outputs as the final content output or manually terminates the process), and
wherein the one or more interaction messages comprise the request from the generative AI model to request the confirmation from the user and an indication of the transmission of the confirmation to the generative AI model (Maschmeyer: [0102]; This process of iteratively modifying input prompts to the generative model and producing new content outputs may be continued until the user either flags one of the outputs as the final content output or manually terminates the process).
It would have been obvious to a person of ordinary skill in the art, before the effective filing date of the invention, to modify Cheng (teaches generating a query-response message pair comprising a query message comprising an action invocation and a response message comprising information responsive to the action invocation; obtaining one or more interaction messages associated with interactions between two or more of a user, an assistant service, and a processing system…transmitting, to the generative AI model, a prompt comprising the first system message, the query-response message pair, the one or more interaction messages, and the second system message; and receiving, from the generative AI model and based at least in part on the prompt, an output of the generative AI model indicating the utterance to be provided to the user by the assistant service) with the teachings of Watson (teaches generating a first system message indicative of a role to be performed by the generative AI model) with the further teachings of Sne (teaches dynamically assembling a prompt comprising the first system message, the query-response message pair, the one or more interaction messages, and the second system message, wherein the first system message and the second system message are positioned to receive greater processing weight from the generative AI model than intermediate messages in the prompt) with the further teachings of Maschmeyer (teaches prompting the user for confirmation of an action selected from the one or more actions available to the generative AI model…to determine whether selected actions are flagged for user confirmation…transmitting the confirmation to the generative AI model, wherein the utterance to be provided to the generative AI model is based at least in part on the confirmation…an indication of the transmission of the confirmation to the generative AI model). One of ordinary skill in the art would have been motivated to make such a combination of evaluating the user input and improving the accuracy of the output based on better prediction (See Maschmeyer: [0049]). In addition, the references (Cheng, Watson, Sne, and Maschmeyer) teach features that are directed to analogous art and they are directed to the same field of endeavor as Cheng, Watson, Sne, and Maschmeyer are directed to natural language conversation system that utilizes assist users accordingly based on the customer requirement.
Regarding claim 13, the modification of Cheng, Watson, and Sne teach the claimed invention substantially as claimed, however the modification of Cheng, Watson, and Sne does not explicitly teach the one or more actions available to the generative AI model comprise prompting the user for confirmation of an action selected from the one or more actions available to the generative AI model; the second system message comprises an instruction for the generative AI model to determine whether selected actions are flagged for user confirmation; and the one or more processors are individually or collectively further operable to execute the code to cause the apparatus to receive a request from the generative AI model to request confirmation of the action from the user, and transmit the confirmation to the generative AI model, wherein the utterance to be provided to the generative AI model is based at least in part on the confirmation, and wherein the one or more interaction messages comprise the request from the generative AI model to request the confirmation from the user and an indication of the transmission of the confirmation to the generative AI model.
Maschmeyer teaches the one or more actions available to the generative AI model comprise prompting the user for confirmation of an action selected from the one or more actions available to the generative AI model (Maschmeyer: [0102]: In particular, the second output may be an output of the generative model that includes representations of user-selected portions from the outputs of preceding iteration(s) of content generation. The second output data may be presented to the user via the user interface. In some implementations, the computing system may prompt the user to confirm the second output data as the final content output or to make further selections of portions from the second output(s));
the second system message comprises an instruction for the generative AI model to determine whether selected actions are flagged for user confirmation (Maschmeyer: [0102]: In particular, the second output may be an output of the generative model that includes representations of user-selected portions from the outputs of preceding iteration(s) of content generation. The second output data may be presented to the user via the user interface. In some implementations, the computing system may prompt the user to confirm the second output data as the final content output or to make further selections of portions from the second output(s). This process of iteratively modifying input prompts to the generative model and producing new content outputs may be continued until the user either flags one of the outputs as the final content output or manually terminates the process); and
the one or more processors are individually or collectively further operable to execute the code to cause the apparatus to receive a request from the generative AI model to request confirmation of the action from the user, and transmit the confirmation to the generative AI model, wherein the utterance to be provided to the generative AI model is based at least in part on the confirmation (Maschmeyer: [0102]; This process of iteratively modifying input prompts to the generative model and producing new content outputs may be continued until the user either flags one of the outputs as the final content output or manually terminates the process), and wherein
the one or more interaction messages comprise the request from the generative AI model to request the confirmation from the user and an indication of the transmission of the confirmation to the generative AI model (Maschmeyer: [0102]; This process of iteratively modifying input prompts to the generative model and producing new content outputs may be continued until the user either flags one of the outputs as the final content output or manually terminates the process).
It would have been obvious to a person of ordinary skill in the art, before the effective filing date of the invention, to modify Cheng (teaches generating a query-response message pair comprising a query message comprising an action invocation and a response message comprising information responsive to the action invocation; obtaining one or more interaction messages associated with interactions between two or more of a user, an assistant service, and a processing system…transmitting, to the generative AI model, a prompt comprising the first system message, the query-response message pair, the one or more interaction messages, and the second system message; and receiving, from the generative AI model and based at least in part on the prompt, an output of the generative AI model indicating the utterance to be provided to the user by the assistant service) with the teachings of Watson (teaches generating a first system message indicative of a role to be performed by the generative AI model) with the further teachings of Sne (teaches dynamically assembling a prompt comprising the first system message, the query-response message pair, the one or more interaction messages, and the second system message, wherein the first system message and the second system message are positioned to receive greater processing weight from the generative AI model than intermediate messages in the prompt) with the further teachings of Maschmeyer (teaches prompting the user for confirmation of an action selected from the one or more actions available to the generative AI model…to determine whether selected actions are flagged for user confirmation…transmitting the confirmation to the generative AI model, wherein the utterance to be provided to the generative AI model is based at least in part on the confirmation…an indication of the transmission of the confirmation to the generative AI model). One of ordinary skill in the art would have been motivated to make such a combination of evaluating the user input and improving the accuracy of the output based on better prediction (See Maschmeyer: [0049]). In addition, the references (Cheng, Watson, Sne, and Maschmeyer) teach features that are directed to analogous art and they are directed to the same field of endeavor as Cheng, Watson, Sne, and Maschmeyer are directed to natural language conversation system that utilizes assist users accordingly based on the customer requirement.
Claims 6 and 15 are rejected under 35 U.S.C. 103 as being unpatentable over U.S Patent Application Publication 2020/0395007 issued to Cheng et al. (hereinafter as "Cheng") in view of U.S Patent 11,971,914 issued to Watson et al. (hereinafter as “Watson”) in view of U.S Patent Application Publication 2025/0139263 issued to Sne et al. (hereinafter as “Sne”) in further view of U.S Patent Application Publication 2024/0356871 issued to Brewer et al. (hereinafter as "Brewer").
Regarding claim 6, the modification of Cheng, Watson, and Sne teach the claimed invention substantially as claimed, however the modification of Cheng, Watson, and Sne does not explicitly teach the first system message is further indicative of a name associated with the generative AI model, an author associated with the generative AI model, a version associated with the generative AI model, a company associated with the generative AI model, a tone associated with the generative AI model, a presence associated with the generative AI model, or any combination thereof.
Brewer teaches the first system message is further indicative of a name associated with the generative AI model, an author associated with the generative AI model, a version associated with the generative AI model (Brewer: [0111]; In some examples, the chatbot backend 430 uses the short-term storage to store group chat session data such as, but not limited to: a user identification of each user in the group chat session; each received and sent message; a timestamp for each message; metadata about which model, parameters where used for generating the response; and keywords that are extracted from each message which could aid in determining conversation context), a company associated with the generative AI model, a tone associated with the generative AI model (Brewer: [0073]; In operation 302, the chatbot system 300 receives, from a user system 314, a prompt 310 of the user 342 during an interactive session. [0083]; In some examples, the chatbot system 300 stores a conversation state in a user profile 322 as part of user database of a series of interactive sessions so that the chatbot system 300 can have a context for conversations that occur over a plurality of interactive sessions. [0089]; In some examples, the chatbot system 300 determines a personality or tone for the responses generated by the chatbot system 300. The chatbot system 300 uses the conversation state and demographic or other information about the user 342 to determine a personality or tone for the user 342, such as by adopting a formal tone for an older user, and a more informal tone for a younger user), a presence associated with the generative AI model, or any combination thereof.
It would have been obvious to a person of ordinary skill in the art, before the effective filing date of the invention, to modify Cheng (teaches generating a first system message indicative of a role to be performed by the generative AI model; generating a query-response message pair comprising a query message comprising an action invocation and a response message comprising information responsive to the action invocation; obtaining one or more interaction messages associated with interactions between two or more of a user, an assistant service, and a processing system…transmitting, to the generative AI model, a prompt comprising the first system message, the query-response message pair, the one or more interaction messages, and the second system message; and receiving, from the generative AI model and based at least in part on the prompt, an output of the generative AI model indicating the utterance to be provided to the user by the assistant service) with the teachings of Watson (teaches generating a first system message indicative of a role to be performed by the generative AI model) with the teachings of Sne (teaches dynamically assembling a prompt comprising the first system message, the query-response message pair, the one or more interaction messages, and the second system message, wherein the first system message and the second system message are positioned to receive greater processing weight from the generative AI model than intermediate messages in the prompt) with the further teachings of Brewer (teaches the first system message is further indicative of a name associated with the generative AI model, an author associated with the generative AI model, a version associated with the generative AI model, a company associated with the generative AI model, a tone associated with the generative AI model, a presence associated with the generative AI model, or any combination thereof). One of ordinary skill in the art would have been motivated to make such a combination of evaluating the user input based on determining the natural language system and the responses to improve the suggestion provided by the chatbot system to improve the user engagement (See Brewer: [0097]). In addition, the references (Cheng, Watson, Sne, and Brewer) teach features that are directed to analogous art and they are directed to the same field of endeavor as Cheng, Watson, Sne, and Brewer are directed to natural language conversation system that utilizes assist users accordingly based on the customer requirement.
Regarding claim 15, the modification of Cheng, Watson, and Sne teach the claimed invention substantially as claimed, however the modification of Cheng, Watson, and Sne does not explicitly teach the first system message is further indicative of a name associated with the generative AI model, an author associated with the generative AI model, a version associated with the generative AI model, a company associated with the generative AI model, a tone associated with the generative AI model, a presence associated with the generative AI model, or any combination thereof.
Brewer teaches the first system message is further indicative of a name associated with the generative AI model, an author associated with the generative AI model, a version associated with the generative AI model (Brewer: [0111]; In some examples, the chatbot backend 430 uses the short-term storage to store group chat session data such as, but not limited to: a user identification of each user in the group chat session; each received and sent message; a timestamp for each message; metadata about which model, parameters where used for generating the response; and keywords that are extracted from each message which could aid in determining conversation context), a company associated with the generative AI model, a tone associated with the generative AI model (Brewer: [0073]; In operation 302, the chatbot system 300 receives, from a user system 314, a prompt 310 of the user 342 during an interactive session. [0083]; In some examples, the chatbot system 300 stores a conversation state in a user profile 322 as part of user database of a series of interactive sessions so that the chatbot system 300 can have a context for conversations that occur over a plurality of interactive sessions. [0089]; In some examples, the chatbot system 300 determines a personality or tone for the responses generated by the chatbot system 300. The chatbot system 300 uses the conversation state and demographic or other information about the user 342 to determine a personality or tone for the user 342, such as by adopting a formal tone for an older user, and a more informal tone for a younger user), a presence associated with the generative AI model, or any combination thereof.
It would have been obvious to a person of ordinary skill in the art, before the effective filing date of the invention, to modify Cheng (teaches generating a first system message indicative of a role to be performed by the generative AI model; generating a query-response message pair comprising a query message comprising an action invocation and a response message comprising information responsive to the action invocation; obtaining one or more interaction messages associated with interactions between two or more of a user, an assistant service, and a processing system…transmitting, to the generative AI model, a prompt comprising the first system message, the query-response message pair, the one or more interaction messages, and the second system message; and receiving, from the generative AI model and based at least in part on the prompt, an output of the generative AI model indicating the utterance to be provided to the user by the assistant service) with the teachings of Watson (teaches generating a first system message indicative of a role to be performed by the generative AI model) with the teachings of Sne (teaches dynamically assembling a prompt comprising the first system message, the query-response message pair, the one or more interaction messages, and the second system message, wherein the first system message and the second system message are positioned to receive greater processing weight from the generative AI model than intermediate messages in the prompt) with the further teachings of Brewer (teaches the first system message is further indicative of a name associated with the generative AI model, an author associated with the generative AI model, a version associated with the generative AI model, a company associated with the generative AI model, a tone associated with the generative AI model, a presence associated with the generative AI model, or any combination thereof). One of ordinary skill in the art would have been motivated to make such a combination of evaluating the user input based on determining the natural language system and the responses to improve the suggestion provided by the chatbot system to improve the user engagement (See Brewer: [0097]). In addition, the references (Cheng, Watson, Sne, and Brewer) teach features that are directed to analogous art and they are directed to the same field of endeavor as Cheng, Watson, Sne, and Brewer are directed to natural language conversation system that utilizes assist users accordingly based on the customer requirement.
Claims 5, 14, and 20 are rejected under 35 U.S.C. 103 as being unpatentable over U.S Patent Application Publication 2020/0395007 issued to Cheng et al. (hereinafter as "Cheng") in view of U.S Patent 11,971,914 issued to Watson et al. (hereinafter as “Watson”) in view of U.S Patent Application Publication 2025/0139263 issued to Sne et al. (hereinafter as “Sne”) in further view of U.S Patent Application Publication 2024/0386213 issued to Ghoche et al. (hereinafter as "Ghoche").
Regarding claim 5, the modification of Cheng, Watson, and Sne teach the claimed invention substantially as claimed, however the modification of Cheng, Watson, and Sne does not explicitly teach the second system message further comprises an indication of one or more response domains selected from a set of response domains comprising a knowledge domain, a frequency asked questions domain, a security quarantine domain, an escalation domain, or any combination thereof; and the one or more response domains are selected based at least in part on an analysis of the first system message, the one or more interaction messages, the second system message, or any combination thereof.
Ghoche teaches the second system message further comprises an indication of one or more response domains selected from a set of response domains comprising a knowledge domain, a frequency asked questions domain, a security quarantine domain, an escalation domain, or any combination thereof (Ghoche: [0073]; The AI augmented customer support module 140 may support one or more functions, such as 1) automatically solving at least a portion of routine customer service support questions; 2) aiding in automatically routing customer service tickets to individual agents, which may include performing a form of triage in which customer tickets in danger of escalation are identified for special service (e.g., to a manager or someone with training in handling escalations); and 3) assisting human agents to formulate responses to complicated questions by, for example, providing suggestions or examples a human agent may select and/or customize. [0175]; Domain relevance is also important. For example, to correctly respond to a customer inquiry about a refund or an order status for a specific business may require understanding context about an inquiry, understanding decision logic, and understanding domain-relevance. [0193]; Empathy customization may include, as inputs, the template answer, the customer question, and the detected topic. These inputs may be provided to a generative model train on empathic conversations and fine tuned to modify an answer to be more empathic in block 3220. Theoretically, other inputs could be provided (e.g., the location of the customer, information on previous interactions with the customer, sentiment/stress metrics based on voice analysis, use of language, or other metrics, etc); and
the one or more response domains are selected based at least in part on an analysis of the first system message, the one or more interaction messages, the second system message, or any combination thereof (Ghoche: [0175]; Domain relevance is also important. For example, to correctly respond to a customer inquiry about a refund or an order status for a specific business may require understanding context about an inquiry, understanding decision logic, and understanding domain-relevance. [0235]; in one implementation, an administrator inputs a natural language description of a workflow policy into a text box 5102. An administrator may also select actions. As examples of actions, FIG. 51 shows a handoff to agent action and an API call (e.g., check order status) in block 5104, although more general a UI may provide a list of all available actions from which to select from. [0238]; It will be understood in the previous examples of the autonomous AI chatbot agent, that it may leverage off previously described techniques for generating a granular taxonomy and determining an intent of a conversation. That is, it may be used in combinations and subcombinations with previously described examples).
It would have been obvious to a person of ordinary skill in the art, before the effective filing date of the invention, to modify Cheng (teaches generating a first system message indicative of a role to be performed by the generative AI model; generating a query-response message pair comprising a query message comprising an action invocation and a response message comprising information responsive to the action invocation; obtaining one or more interaction messages associated with interactions between two or more of a user, an assistant service, and a processing system…transmitting, to the generative AI model, a prompt comprising the first system message, the query-response message pair, the one or more interaction messages, and the second system message; and receiving, from the generative AI model and based at least in part on the prompt, an output of the generative AI model indicating the utterance to be provided to the user by the assistant service) with the teachings of Watson (teaches generating a first system message indicative of a role to be performed by the generative AI model) with the teachings of Sne (teaches dynamically assembling a prompt comprising the first system message, the query-response message pair, the one or more interaction messages, and the second system message, wherein the first system message and the second system message are positioned to receive greater processing weight from the generative AI model than intermediate messages in the prompt) with the further teachings of Ghoche (teaches the second system message further comprises an indication of one or more response domains selected from a set of response domains comprising a knowledge domain, a frequency asked questions domain, a security quarantine domain, an escalation domain, or any combination thereof; and the one or more response domains are selected based at least in part on an analysis of the first system message, the one or more interaction messages, the second system message, or any combination thereof). One of ordinary skill in the art would have been motivated to make such a combination of evaluating the user input based on determining the natural language system and the responses to improve the insight in servicing and assist customers accordingly (See Ghoche [0119]). In addition, the references (Cheng, Watson, Sne, and Ghoche) teach features that are directed to analogous art and they are directed to the same field of endeavor as Cheng, Watson, Sne, and Ghoche are directed to natural language conversation system that utilizes assist users accordingly based on the customer requirement.
Regarding claim 14, the modification of Cheng, Watson, and Sne teach the claimed invention substantially as claimed, however the modification of Cheng, Watson, and Sne does not explicitly teach the second system message further comprises an indication of one or more response domains selected from a set of response domains comprising a knowledge domain, a frequency asked questions domain, a security quarantine domain, an escalation domain, or any combination thereof.
Ghoche teaches the second system message further comprises an indication of one or more response domains selected from a set of response domains comprising a knowledge domain, a frequency asked questions domain, a security quarantine domain, an escalation domain, or any combination thereof (Ghoche: [0073]; The AI augmented customer support module 140 may support one or more functions, such as 1) automatically solving at least a portion of routine customer service support questions; 2) aiding in automatically routing customer service tickets to individual agents, which may include performing a form of triage in which customer tickets in danger of escalation are identified for special service (e.g., to a manager or someone with training in handling escalations); and 3) assisting human agents to formulate responses to complicated questions by, for example, providing suggestions or examples a human agent may select and/or customize. [0175]; Domain relevance is also important. For example, to correctly respond to a customer inquiry about a refund or an order status for a specific business may require understanding context about an inquiry, understanding decision logic, and understanding domain-relevance. [0193]; Empathy customization may include, as inputs, the template answer, the customer question, and the detected topic. These inputs may be provided to a generative model train on empathic conversations and fine tuned to modify an answer to be more empathic in block 3220. Theoretically, other inputs could be provided (e.g., the location of the customer, information on previous interactions with the customer, sentiment/stress metrics based on voice analysis, use of language, or other metrics, etc); and
the one or more response domains are selected based at least in part on an analysis of the first system message, the one or more interaction messages, the second system message, or any combination thereof (Ghoche: [0175]; Domain relevance is also important. For example, to correctly respond to a customer inquiry about a refund or an order status for a specific business may require understanding context about an inquiry, understanding decision logic, and understanding domain-relevance. [0235]; in one implementation, an administrator inputs a natural language description of a workflow policy into a text box 5102. An administrator may also select actions. As examples of actions, FIG. 51 shows a handoff to agent action and an API call (e.g., check order status) in block 5104, although more general a UI may provide a list of all available actions from which to select from. [0238]; It will be understood in the previous examples of the autonomous AI chatbot agent, that it may leverage off previously described techniques for generating a granular taxonomy and determining an intent of a conversation. That is, it may be used in combinations and subcombinations with previously described examples).
It would have been obvious to a person of ordinary skill in the art, before the effective filing date of the invention, to modify Cheng (teaches generating a first system message indicative of a role to be performed by the generative AI model; generating a query-response message pair comprising a query message comprising an action invocation and a response message comprising information responsive to the action invocation; obtaining one or more interaction messages associated with interactions between two or more of a user, an assistant service, and a processing system…transmitting, to the generative AI model, a prompt comprising the first system message, the query-response message pair, the one or more interaction messages, and the second system message; and receiving, from the generative AI model and based at least in part on the prompt, an output of the generative AI model indicating the utterance to be provided to the user by the assistant service) with the teachings of Watson (teaches generating a first system message indicative of a role to be performed by the generative AI model) with the teachings of Sne (teaches dynamically assembling a prompt comprising the first system message, the query-response message pair, the one or more interaction messages, and the second system message, wherein the first system message and the second system message are positioned to receive greater processing weight from the generative AI model than intermediate messages in the prompt) with the further teachings of Ghoche (teaches the second system message further comprises an indication of one or more response domains selected from a set of response domains comprising a knowledge domain, a frequency asked questions domain, a security quarantine domain, an escalation domain, or any combination thereof; and the one or more response domains are selected based at least in part on an analysis of the first system message, the one or more interaction messages, the second system message, or any combination thereof). One of ordinary skill in the art would have been motivated to make such a combination of evaluating the user input based on determining the natural language system and the responses to improve the insight in servicing and assist customers accordingly (See Ghoche [0119]). In addition, the references (Cheng, Watson, Sne, and Ghoche) teach features that are directed to analogous art and they are directed to the same field of endeavor as Cheng, Watson, Sne, and Ghoche are directed to natural language conversation system that utilizes assist users accordingly based on the customer requirement.
Regarding claim 20, the modification of Cheng, Watson, and Sne teaches the claimed invention substantially as claimed, however the modification of Cheng, Watson, and Sne does not explicitly teach the second system message further comprises an indication of one or more response domains selected from a set of response domains comprising a knowledge domain, a frequency asked questions domain, a security quarantine domain, an escalation domain, or any combination thereof; and the one or more response domains are selected based at least in part on an analysis of the first system message, the one or more interaction messages, the second system message, or any combination thereof.
Ghoche teaches the second system message further comprises an indication of one or more response domains selected from a set of response domains comprising a knowledge domain, a frequency asked questions domain, a security quarantine domain, an escalation domain, or any combination thereof (Ghoche: [0073]; The AI augmented customer support module 140 may support one or more functions, such as 1) automatically solving at least a portion of routine customer service support questions; 2) aiding in automatically routing customer service tickets to individual agents, which may include performing a form of triage in which customer tickets in danger of escalation are identified for special service (e.g., to a manager or someone with training in handling escalations); and 3) assisting human agents to formulate responses to complicated questions by, for example, providing suggestions or examples a human agent may select and/or customize. [0175]; Domain relevance is also important. For example, to correctly respond to a customer inquiry about a refund or an order status for a specific business may require understanding context about an inquiry, understanding decision logic, and understanding domain-relevance. [0193]; Empathy customization may include, as inputs, the template answer, the customer question, and the detected topic. These inputs may be provided to a generative model train on empathic conversations and fine tuned to modify an answer to be more empathic in block 3220. Theoretically, other inputs could be provided (e.g., the location of the customer, information on previous interactions with the customer, sentiment/stress metrics based on voice analysis, use of language, or other metrics, etc); and
the one or more response domains are selected based at least in part on an analysis of the first system message, the one or more interaction messages, the second system message, or any combination thereof (Ghoche: [0175]; Domain relevance is also important. For example, to correctly respond to a customer inquiry about a refund or an order status for a specific business may require understanding context about an inquiry, understanding decision logic, and understanding domain-relevance. [0235]; in one implementation, an administrator inputs a natural language description of a workflow policy into a text box 5102. An administrator may also select actions. As examples of actions, FIG. 51 shows a handoff to agent action and an API call (e.g., check order status) in block 5104, although more general a UI may provide a list of all available actions from which to select from. [0238]; It will be understood in the previous examples of the autonomous AI chatbot agent, that it may leverage off previously described techniques for generating a granular taxonomy and determining an intent of a conversation. That is, it may be used in combinations and subcombinations with previously described examples).
It would have been obvious to a person of ordinary skill in the art, before the effective filing date of the invention, to modify Cheng (teaches generating a first system message indicative of a role to be performed by the generative AI model; generating a query-response message pair comprising a query message comprising an action invocation and a response message comprising information responsive to the action invocation; obtaining one or more interaction messages associated with interactions between two or more of a user, an assistant service, and a processing system…transmitting, to the generative AI model, a prompt comprising the first system message, the query-response message pair, the one or more interaction messages, and the second system message; and receiving, from the generative AI model and based at least in part on the prompt, an output of the generative AI model indicating the utterance to be provided to the user by the assistant service) with the teachings of Watson (teaches generating a first system message indicative of a role to be performed by the generative AI model) with the teachings of Sne (teaches dynamically assembling a prompt comprising the first system message, the query-response message pair, the one or more interaction messages, and the second system message, wherein the first system message and the second system message are positioned to receive greater processing weight from the generative AI model than intermediate messages in the prompt) with the further teachings of Ghoche (teaches the second system message further comprises an indication of one or more response domains selected from a set of response domains comprising a knowledge domain, a frequency asked questions domain, a security quarantine domain, an escalation domain, or any combination thereof; and the one or more response domains are selected based at least in part on an analysis of the first system message, the one or more interaction messages, the second system message, or any combination thereof). One of ordinary skill in the art would have been motivated to make such a combination of evaluating the user input based on determining the natural language system and the responses to improve the insight in servicing and assist customers accordingly (See Ghoche [0119]). In addition, the references (Cheng, Watson, Sne, and Ghoche) teach features that are directed to analogous art and they are directed to the same field of endeavor as Cheng, Watson, Sne, and Ghoche are directed to natural language conversation system that utilizes assist users accordingly based on the customer requirement.
Claims 8 and 17 are rejected under 35 U.S.C. 103 as being unpatentable over U.S Patent Application Publication 2020/0395007 issued to Cheng et al. (hereinafter as "Cheng") in view of U.S Patent 11,971,914 issued to Watson et al. (hereinafter as “Watson”) in view of U.S Patent Application Publication 2025/0139263 issued to Sne et al. (hereinafter as “Sne”) in further view of U.S Patent Application Publication 2023/0409836 issued to Balasubrhamania et al. (hereinafter as “Balasubrhamania”).
Regarding claim 8, the modification of Cheng, Watson, and Sne teaches the claimed invention substantially as claimed, however the modification of Cheng, Watson, and Sne does not explicitly teach validating the output of the generative AI model to determine whether the utterance indicated in the output of the generative AI model is responsive to the instruction for the generative AI model to generate the utterance.
Balasubrhamania teaches validating the output of the generative AI model to determine whether the utterance indicated in the output of the generative AI model is responsive to the instruction for the generative AI model to generate the utterance (Balasubrhamania: [0054]; The conversation handler 230 can distinguish between and select from among these different ways to construct a query based on the structural entities returned by the semantic processor and/or the way that the user request is presented. [0083]-[0084]; a text block 655 is reached that indicates that the user should provide more information, for example: “Please provide more details.” In one embodiment, the conversation handler 230 returns this text to the user interaction interface 220 for presentation (by text, voice, and/or GUI) to the user to prompt the user to provide more information. Where the system is unable to resolve whether there is or is not context and/or keyword(s) provided sufficient to identify records (proceed from block 605 to block 620), an error message block 660 is reached that indicates that the request cannot be processed. For example, the error message may indicate “Sorry, I cannot handle the request.” In one embodiment, the conversation handler 230 returns this error message to the user interaction interface 220 for presentation (by text, voice, and/or GUI)).
It would have been obvious to a person of ordinary skill in the art, before the effective filing date of the invention, to modify Cheng (teaches generating a query-response message pair comprising a query message comprising an action invocation and a response message comprising information responsive to the action invocation; obtaining one or more interaction messages associated with interactions between two or more of a user, an assistant service, and a processing system…transmitting, to the generative AI model, a prompt comprising the first system message, the query-response message pair, the one or more interaction messages, and the second system message; and receiving, from the generative AI model and based at least in part on the prompt, an output of the generative AI model indicating the utterance to be provided to the user by the assistant service) with the teachings of Watson (teaches generating a first system message indicative of a role to be performed by the generative AI model) with the teachings of Sne (teaches dynamically assembling a prompt comprising the first system message, the query-response message pair, the one or more interaction messages, and the second system message, wherein the first system message and the second system message are positioned to receive greater processing weight from the generative AI model than intermediate messages in the prompt) with the further teachings of Balasubrhamania (teaches validating the output of the generative AI model to determine whether the utterance indicated in the output of the generative AI model is responsive to the instruction for the generative AI model to generate the utterance). One of ordinary skill in the art would have been motivated to make such a combination of improving the conversational interfaces by permitting changes to the construction of the query on the fly (See Balasubrhamania: [0023]). In addition, the references (Cheng, Watson, Sne, and Balasubrhamania) teach features that are directed to analogous art and they are directed to the same field of endeavor as Cheng, Watson, Sne, and Balasubrhamania are directed to natural language conversation system that utilizes assist users accordingly based on the customer requirement.
Regarding claim 17, the modification of Cheng, Watson, and Sne teaches the claimed invention substantially as claimed, however the modification of Cheng, Watson, and Sne does not explicitly teach the one or more processors are individually or collectively further operable to execute the code to cause the apparatus to: validate the output of the generative AI model to determine whether the utterance indicated in the output of the generative AI model is responsive to the instruction for the generative AI model to generate the utterance.
Balasubrhamania teaches the one or more processors are individually or collectively further operable to execute the code to cause the apparatus to: validate the output of the generative AI model to determine whether the utterance indicated in the output of the generative AI model is responsive to the instruction for the generative AI model to generate the utterance (Balasubrhamania: [0054]; The conversation handler 230 can distinguish between and select from among these different ways to construct a query based on the structural entities returned by the semantic processor and/or the way that the user request is presented. [0083]-[0084]; a text block 655 is reached that indicates that the user should provide more information, for example: “Please provide more details.” In one embodiment, the conversation handler 230 returns this text to the user interaction interface 220 for presentation (by text, voice, and/or GUI) to the user to prompt the user to provide more information. Where the system is unable to resolve whether there is or is not context and/or keyword(s) provided sufficient to identify records (proceed from block 605 to block 620), an error message block 660 is reached that indicates that the request cannot be processed. For example, the error message may indicate “Sorry, I cannot handle the request.” In one embodiment, the conversation handler 230 returns this error message to the user interaction interface 220 for presentation (by text, voice, and/or GUI)).
It would have been obvious to a person of ordinary skill in the art, before the effective filing date of the invention, to modify Cheng (teaches generating a query-response message pair comprising a query message comprising an action invocation and a response message comprising information responsive to the action invocation; obtaining one or more interaction messages associated with interactions between two or more of a user, an assistant service, and a processing system…transmitting, to the generative AI model, a prompt comprising the first system message, the query-response message pair, the one or more interaction messages, and the second system message; and receiving, from the generative AI model and based at least in part on the prompt, an output of the generative AI model indicating the utterance to be provided to the user by the assistant service) with the teachings of Watson (teaches generating a first system message indicative of a role to be performed by the generative AI model) with the teachings of Sne (teaches dynamically assembling a prompt comprising the first system message, the query-response message pair, the one or more interaction messages, and the second system message, wherein the first system message and the second system message are positioned to receive greater processing weight from the generative AI model than intermediate messages in the prompt) with the further teachings of Balasubrhamania (teaches validating the output of the generative AI model to determine whether the utterance indicated in the output of the generative AI model is responsive to the instruction for the generative AI model to generate the utterance). One of ordinary skill in the art would have been motivated to make such a combination of improving the conversational interfaces by permitting changes to the construction of the query on the fly (See Balasubrhamania: [0023]). In addition, the references (Cheng, Watson, Sne, and Balasubrhamania) teach features that are directed to analogous art and they are directed to the same field of endeavor as Cheng, Watson, Sne, and Balasubrhamania are directed to natural language conversation system that utilizes assist users accordingly based on the customer requirement.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
U.S Patent 11,868,384 issued to Shah et al. (hereinafter as “Shah”) teaches automatically determining responses in a messaging platform that includes receiving a set of inputs and providing the set of responses option based on the user.
U.S Patent 11,947,902 issued to Grimshaw et al. (hereinafter as “Grimshaw”) teaches using a generative artificial intelligence model using mult-turn process to generate a suggested draft replay to a selected message.
U.S Patent 11,960,514 issued to Taylert et al. (hereinafter as “Taylert”) teaches generating content in association with an information search and retrieval system by generating a modified information and passed it back to the user.
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.
Contact Information
Any inquiry concerning this communication or earlier communications from the examiner should be directed to ANDREW N HO whose telephone number is (571)270-0590. The examiner can normally be reached Tuesday and Thursday 10:00-6:00.
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, Sherief Badawi can be reached at (571) 272-9782. 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.
4/24/2026
/ANDREW N HO/Examiner
Art Unit 2169
/SHERIEF BADAWI/Supervisory Patent Examiner, Art Unit 2169
Prosecution Insights