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 .
Introduction
The following is a non-final Office Action in response to Applicant’s submission filed on 6/30/2025. Currently claims 1-20 are pending and claims 1, 11, 20 are independent.
Priority
Acknowledgment is made of applicant’s claim for foreign priority under 35 U.S.C. 119 (a)-(d). The certified copy has been filed in parent Application No. CN202411448809.0, filed on 10/16/2024.
Information Disclosure Statement
The information disclosure statement (IDS) submitted on 6/30/2025 and 2/23/2026 appears to be in compliance with the provisions of 37 CFR 1.97. Accordingly, the IDS is being considered by the Examiner.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to a judicial exception (i.e., a law of nature, a natural phenomenon, or an abstract idea), specifically an abstract idea, without significantly more. With respect to claims 1-20, following the guidance contained within MPEP 2106, the inquiry for patent eligibility follows two steps: Step 1: Does the claimed invention fall within one of the four statutory categories of invention? Step 2A (Prong 1): Is the claim “directed to” an abstract idea? Step 2A (Prong 2): Is the claim integrated into a practical application? Step 2B: Does the claim recite additional elements that amount to “significantly more” than the abstract idea?
In accordance with these steps, the Examiner finds the following:
Step 1: Claim 1 and its dependent claims (claims 2-10) are directed to a statutory category, namely a method. Claim 11 and its dependent claims (claims 12-19) are directed to a statutory category, namely a system/machine. Claim 20 is directed to a statutory category, namely an article of manufacture.
Step 2A (Prong 1): Claims 1, 11, and 20, which are substantially similar claims to one another, are directed to the abstract idea of “Certain methods of organizing human activity”, or more particularly, “Concepts relating to commercial or legal interactions (including: advertising, marketing or sales activities or behaviors; business relations) (See MPEP 2106).” In this application that refers to using a computer system to determine the best assistant for completing a task. To clarify this further, the Applicant’s disclosed invention is a conceptual system meant to perform the same function that a customer service call center performs when routing and answering calls. The abstract elements of claims 1, 11, and 20, recite in part “Receive input information…Determine target digital assistant…Present target response…”. Dependent claims 2-10, 12-19, add to the abstract idea the following limitations which recite in part “Determine type of target task…Determine target digital assistant…Obtain result…Generate response…Obtain result…Determine different assistant…Recognize target information…Present result…Update result…Present updated result…Replace result…Apply mask…Determine permission…Determine accessible assistant…Determine target assistant…Provide permission information…Receive second information…Present response…”. All of these additional limitations, however, only serve to further limit the abstract idea, and hence are nonetheless directed towards fundamentally the same abstract idea as independent claims 1, 11, and 20.
Step 2A (Prong 2): Independent claims 1, 11, and 20, which are substantially similar claims to one another, do not contain additional elements, either considered individually or in combination, that effectively integrate the exception into a practical application of the exception. These claims do include the limitation that recites in part “Processors…Memory…Device…Interface…Non-transitory computer readable medium…Digital assistant…” which limits the claims to a networked/computer based environment, but this is insufficient with respect to integration into a practical application because it is merely applying the abstract idea to a general computer (See MPEP 2106.05(f)).
Dependent claims 2, 4, 12, 14, add the additional element which recites in part “ML model…” which again limits the claims to a networked/computer based environment, but this is insufficient with respect to integration into a practical application because it is again merely applying the abstract idea to a general computer (See MPEP 2106.05(f)).
Additionally, dependent claims 3, 5-10, 13, 15-19 do not include any additional elements to conduct a further Step 2A (Prong 2) analysis.
Step 2B: Independent claims 1, 11, and 20, which are substantially similar claims to one another, include additional elements, when considered both individually and as an ordered combination, which are insufficient to amount to significantly more than the judicial exception. The additional elements of these claims recite in part “Processors…Memory…Device…Interface…Non-transitory computer readable medium…Digital assistant…”. These items are not significantly more because these are merely the software and/or hardware components used to implement the abstract idea (determine the best assistant for completing a task) on a general purpose computer (See MPEP 2106.05(f)). This is exemplified in the Applicant’s specification in [0099] – “These computer-readable program instructions may be provided to a processing unit of a general- purpose computer.”
Dependent claims 2, 4, 12, 14 include additional elements, when considered both individually and as an ordered combination and in view of their respective independent claims, which are insufficient to amount to significantly more than the judicial exception. Specifically, dependent claims 2, 4, 12, 14 include the additional element which recites in part “ML model…” These are the same additional elements that are addressed above in claims 1, 11, and 20, and are not significantly more because these are merely the software and/or hardware components used to implement the abstract idea (determine the best assistant for completing a task) on a general purpose computer (See MPEP 2106.05(f)).
Additionally, dependent claims3, 5-10, 13, 15-19 do not include any additional elements to conduct a further 2B analysis.
Accordingly, whether taken individually or as an ordered combination claims 1-20 are rejected under 35 USC § 101 because the claimed invention is directed to a judicial exception, an abstract idea, without significantly more.
Claim Rejections - 35 USC § 102
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(2) the claimed invention was described in a patent issued under section 151, or in an application for patent published or deemed published under section 122(b), in which the patent or application, as the case may be, names another inventor and was effectively filed before the effective filing date of the claimed invention.
Claims 1-4, 10-14, 20 are rejected under 35 U.S.C. 102(a)(2) as being anticipated by Zhang et al. (US 20250190449 A1)
Regarding claims 1, 11, 20, Zhang discloses a method for interacting with a digital assistant (Zhang ¶1 - Particular implementations leverage trained artificial intelligence models to determine intent of user prompts and to intelligently select generative artificial intelligence agents to perform analytics tasks to generate responses to the user prompts), comprising: receiving, through a first interaction interface of a first digital assistant, first input information input by a first user (Zhang Fig. 8-9 - Zhang ABS - a system may receive a prompt that includes information associated with a structured data set which includes at least some numerical data); determining, based on the first input information, a target digital assistant for the first input information from a plurality of second digital assistants, different digital assistants in the plurality of second digital assistants configured to perform tasks of different types (Zhang Fig. 8-9 - Zhang ABS - The system may provide the prompt as input to an agent orchestrator to select one or more generative AI agents to perform analytics tasks corresponding to the information); and presenting, in the first interaction interface, target response information for the first input information, the target response information obtained based at least on processing of the first input information by the target digital assistant (Zhang ¶94 - After completion of operations described with reference to 1110-1116, the process flow 1100 progresses to 1118, and a response is generated. For example, the response may include a GUI that provides a combination of text output, numerical output, and visualization elements).
Regarding claims 2, 12, Zhang discloses determining, by using a machine learning model associated with the first digital assistant, a type of a target task indicated by the first input information; and determining, based on the type of the target task, the target digital assistant from at least one second digital assistant in the plurality of second digital assistants (Zhang ¶81 - The agent orchestrator 812 is configured to output an agent assignment 814 that indicates a selection of one or more generative AI agents based on the agent features 806, the intent features 808, and the data schema features 810…In some implementations, the agent orchestrator 812 includes or corresponds to a trained AI or ML classifier that is trained to select generative AI agents based on input features. For example, the agent orchestrator 812 may be trained based on labeled input features that are labeled according to selected generative AI agents using a supervised learning process).
Regarding claims 3, 13, Zhang discloses obtaining a processing result for the first input information provided by the target digital assistant; and generating, by using the first digital assistant, the target response information based on the processing result (Zhang Fig. 11 - Zhang ¶94 - After completion of operations described with reference to 1110-1116, the process flow 1100 progresses to 1118, and a response is generated. For example, the response may include a GUI that provides a combination of text output, numerical output, and visualization elements).
Regarding claims 4, 14, Zhang discloses obtaining a processing result for the first input information and provided by a candidate digital assistant in the plurality of second digital assistants; and determining, by using a machine learning model associated with the first digital assistant and in response to the processing result not matching the first input information or the processing result indicating a processing failure, a digital assistant different from the candidate digital assistant in the plurality of second digital assistants as the target digital assistant (Zhang ¶97 - The method 1200 includes executing an ensemble model to generate a response to the prompt based on the structured data set, at 1206. The ensemble model includes the one or more generative AI agents. For example, the response may include or correspond to the response 110 of FIG. 1, and the ensemble model may include one or more of the generative AI agents 126 of FIG. 1. In some implementations, the ensemble model includes one or more differently selected generative AI agents from the plurality of generative AI agents according to a greedy algorithm parameter).
Regarding claims 10, Zhang discloses receiving, through a second interaction interface of the target digital assistant, second input information input by the first user; and presenting, in the second interaction interface, response information obtained by processing the second input information by the target digital assistant (Zhang ¶90 - In some implementations, the user may provide user feedback 1006 based on the response 1004, and the ensemble model 1000 may update the parameters of the response selection process based on the user feedback 1006 {i.e. second input}. For example, the ensemble model 1000 may adjust one or more of criteria used by the agent ranker 1010 based on the user feedback 1006. As another example, the ensemble model 1000 may adjust one or more of the chain of AI thoughts 1012, or may add an AI thought to or delete an AI thought from the chain of AI thoughts 1012, based on the user feedback 1006. In this manner, the ensemble model 1000 may perform continuous learning to increase accuracy and/or user satisfaction with the generated responses).
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.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claims 5-7, 15-17 are rejected under 35 U.S.C. 103 as being unpatentable over Zhang et al. (US 20250190449 A1) in view of Spencer et al. (US 20250307418 A1)
Regarding claims 5, 15, Zhang discloses a method for interacting with a digital assistant (Zhang ¶1 - Particular implementations leverage trained artificial intelligence models to determine intent of user prompts and to intelligently select generative artificial intelligence agents to perform analytics tasks to generate responses to the user prompts).
Zhang lacks recognizing target information in the processing result for the first input information and provided by the target digital assistant; and presenting, in response to no target information being recognized, the processing result in the first interaction interface as the target response information.
Spencer, from the same field of endeavor, teaches recognizing target information in the processing result for the first input information and provided by the target digital assistant; and presenting, in response to no target information being recognized, the processing result in the first interaction interface as the target response information (Spencer ¶66 - To illustrate, consider a scenario where a large language model (LLM) is used for generating responses to customer queries in a customer service chatbot application. The LLM output inspectors 250, based on the administrative policy, are configured to ensure that the responses generated do not contain any confidential customer information such as credit card numbers or personal addresses. If during the analysis, an LLM output inspector detects such sensitive information in the output data, it triggers a violation of the LLM output setting in the administrative policy. Subsequently, upon determining that the LLM output data 160 violates the LLM output setting of the administrative policy, the present technology blocks the LLM output data 160 from being transmitted to the end-user or client application).
It would be obvious for one of ordinary skill in the art before the effective filing date of the Applicant’s claimed invention to modify the generative AI analytics methodology/system of Zhang by including the AI model guardrail techniques of Spencer because Spencer discloses “this dynamic routing capability ensures efficient utilization of resources and optimal performance of the system in handling diverse types of LLM input data (Spencer ¶74)”. Additionally, Zhang further details that it “the agent orchestrator includes a trained AI classifier configured to select the one or more generative AI agents from a plurality of generative AI agents (Zhang ABS)” so it would be obvious to consider including the additional AI model guardrail techniques that Spencer discloses because it would improve the orchestration of Zhang by enabling efficient dynamic routing of tasks.
Regarding claims 6, 16, Zhang in view of Spencer discloses a method for interacting with a digital assistant (Zhang ¶1 - Particular implementations leverage trained artificial intelligence models to determine intent of user prompts and to intelligently select generative artificial intelligence agents to perform analytics tasks to generate responses to the user prompts).
Spencer further teaches updating, in response to the target information being recognized, the processing result based on a configured blocking policy; and presenting the updated processing result in the first interaction interface as the target response information (Spencer ¶66 - To illustrate, consider a scenario where a large language model (LLM) is used for generating responses to customer queries in a customer service chatbot application. The LLM output inspectors 250, based on the administrative policy, are configured to ensure that the responses generated do not contain any confidential customer information such as credit card numbers or personal addresses. If during the analysis, an LLM output inspector detects such sensitive information in the output data, it triggers a violation of the LLM output setting in the administrative policy. Subsequently, upon determining that the LLM output data 160 violates the LLM output setting of the administrative policy, the present technology blocks the LLM output data 160 from being transmitted to the end-user or client application).
It would be obvious for one of ordinary skill in the art before the effective filing date of the Applicant’s claimed invention to modify the generative AI analytics methodology/system of Zhang by including the AI model guardrail techniques of Spencer because Spencer discloses “this dynamic routing capability ensures efficient utilization of resources and optimal performance of the system in handling diverse types of LLM input data (Spencer ¶74)”. Additionally, Zhang further details that it “the agent orchestrator includes a trained AI classifier configured to select the one or more generative AI agents from a plurality of generative AI agents (Zhang ABS)” so it would be obvious to consider including the additional AI model guardrail techniques that Spencer discloses because it would improve the orchestration of Zhang by enabling efficient dynamic routing of tasks.
Regarding claims 7, 17, Zhang in view of Spencer discloses a method for interacting with a digital assistant (Zhang ¶1 - Particular implementations leverage trained artificial intelligence models to determine intent of user prompts and to intelligently select generative artificial intelligence agents to perform analytics tasks to generate responses to the user prompts).
Spencer further teaches replacing the processing result with preset information, the preset information indicating that the processing result contains the target information and cannot be presented, removing the target information from the processing result, or applying a mask to the target information (Spencer ¶66 - To illustrate, consider a scenario where a large language model (LLM) is used for generating responses to customer queries in a customer service chatbot application. The LLM output inspectors 250, based on the administrative policy, are configured to ensure that the responses generated do not contain any confidential customer information such as credit card numbers or personal addresses. If during the analysis, an LLM output inspector detects such sensitive information in the output data, it triggers a violation of the LLM output setting in the administrative policy. Subsequently, upon determining that the LLM output data 160 violates the LLM output setting of the administrative policy, the present technology blocks the LLM output data 160 from being transmitted to the end-user or client application).
It would be obvious for one of ordinary skill in the art before the effective filing date of the Applicant’s claimed invention to modify the generative AI analytics methodology/system of Zhang by including the AI model guardrail techniques of Spencer because Spencer discloses “this dynamic routing capability ensures efficient utilization of resources and optimal performance of the system in handling diverse types of LLM input data (Spencer ¶74)”. Additionally, Zhang further details that it “the agent orchestrator includes a trained AI classifier configured to select the one or more generative AI agents from a plurality of generative AI agents (Zhang ABS)” so it would be obvious to consider including the additional AI model guardrail techniques that Spencer discloses because it would improve the orchestration of Zhang by enabling efficient dynamic routing of tasks.
Claims 8, 9, 18, 19 are rejected under 35 U.S.C. 103 as being unpatentable over Zhang et al. (US 20250190449 A1) in view of Bell et al. (US 20240404687 A1)
Regarding claims 8, 18, Zhang discloses a method for interacting with a digital assistant (Zhang ¶1 - Particular implementations leverage trained artificial intelligence models to determine intent of user prompts and to intelligently select generative artificial intelligence agents to perform analytics tasks to generate responses to the user prompts).
Zhang lacks determining, based on login information of the first user, permission information of the first user; determining, based on the permission information and from the plurality of second digital assistants, at least one second digital assistant accessible by the first user; and determining, based on the first input information, the target digital assistant from the at least one second digital assistant.
Bell, from the same field of endeavor, teaches determining, based on login information of the first user, permission information of the first user; determining, based on the permission information and from the plurality of second digital assistants, at least one second digital assistant accessible by the first user; and determining, based on the first input information, the target digital assistant from the at least one second digital assistant (Bell ¶275 - The computing system determines (2504) a set of task-specific components (e.g., agent modules) and a set of databases to which the user identifier has access. In some embodiments, the sets of task-specific components and/or databases are determined based on an access level and/or permissions associated with the user identifier. In some embodiments, the databases (and/or data within the databases) are subject to different access control lists. In some of these embodiments, the user identifier is checked against the access control lists to determine what data the user is authorized to access).
It would be obvious for one of ordinary skill in the art before the effective filing date of the Applicant’s claimed invention to modify the generative AI analytics methodology/system of Zhang by including the AI agent routing techniques of Bell because Bell discloses “Identifying the correct agent type for a given task can improve performance and reduce computer and storage costs (Bell ¶10)”. Additionally, Zhang further details that it “the agent orchestrator includes a trained AI classifier configured to select the one or more generative AI agents from a plurality of generative AI agents (Zhang ABS)” so it would be obvious to consider including the additional AI agent routing techniques that Bell discloses because it would improve the orchestration of Zhang by enabling correct assignment of tasks.
Regarding claims 9, 19, Zhang in view of Bell discloses a method for interacting with a digital assistant (Zhang ¶1 - Particular implementations leverage trained artificial intelligence models to determine intent of user prompts and to intelligently select generative artificial intelligence agents to perform analytics tasks to generate responses to the user prompts).
Bell further teaches providing the permission information to the target digital assistant, to enable the target digital assistant to determine, based on the permission information, a knowledge base or a function module accessible by the first user (Bell ¶275 - The computing system determines (2504) a set of task-specific components (e.g., agent modules) and a set of databases to which the user identifier has access. In some embodiments, the sets of task-specific components and/or databases are determined based on an access level and/or permissions associated with the user identifier. In some embodiments, the databases (and/or data within the databases) are subject to different access control lists. In some of these embodiments, the user identifier is checked against the access control lists to determine what data the user is authorized to access).
It would be obvious for one of ordinary skill in the art before the effective filing date of the Applicant’s claimed invention to modify the generative AI analytics methodology/system of Zhang by including the AI agent routing techniques of Bell because Bell discloses “Identifying the correct agent type for a given task can improve performance and reduce computer and storage costs (Bell ¶10)”. Additionally, Zhang further details that it “the agent orchestrator includes a trained AI classifier configured to select the one or more generative AI agents from a plurality of generative AI agents (Zhang ABS)” so it would be obvious to consider including the additional AI agent routing techniques that Bell discloses because it would improve the orchestration of Zhang by enabling correct assignment of tasks.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure:
Peng (US 20260086825 A1)
Chen et al. (US 10854203 B2)
Yu et al. (CN 118535688 A)
and
Clarke, Christopher, et al. "One Agent Too Many: User Perspectives on Approaches to Multi-agent Conversational AI." arXiv preprint arXiv:2401.07123 (2024) [online], [retrieved on 2026-07-11]. Retrieved from the Internet <https://arxiv.org/abs/2401.07123 >
These pieces of prior art are cited because they disclose variations on agent selection and orchestration.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Michael R Koester whose telephone number is (313)446-4837. The examiner can normally be reached Monday thru Friday 8:00AM-5:00 PM EST.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Jerry O'Connor can be reached at (571) 272-6787. 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.
/MICHAEL R KOESTER/Examiner, Art Unit 3624
/Jerry O'Connor/Supervisory Patent Examiner,Group Art Unit 3624
Prosecution Insights