DETAILED ACTION
Application No. 19/075,141 filed on 03/10/2025 has been examined. The response to Election Requirement filed on 05/07/2026 has been entered. Group 1 (claims 1-9) has been elected.
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 .
Information Disclosure Statement
The information disclosure statement (IDS) submitted on 03/24/2025, 10/02/2025 and 10/03/2025 is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement 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 1-9 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) without significantly more.
Based upon consideration of all of the relevant factors with respect to the claims as a whole, claims 1-9 are determined to be directed to an abstract idea and not significantly more than the abstract idea itself. The rationale for this determination is explained below:
At Step 1:
Independent claim 1 recites:
“a system for facilitating single point of user access to multiple large language models to generate a response to a user query from an access, the system comprising: a first network of agents, wherein at least two agents facilitate access to at least two different large language models; and an interface for receiving a user's query and initiating a session with the first network of agents, wherein initiating the session includes creating an instance of a front man agent having access to a first large language model and further wherein the front man agent has access to at least a first branch agent, the first branch agent having access to a second large language model.”
At Step 2A, Prong One:
The claim recites limitations directed to an abstract idea.
The limitation of “an interface for receiving a user’s query”, as drafted, is a process that, under the broadest reasonable interpretation, covers a mental process because a human mind can receive and understand a query by reading or hearing the query.
For example, a person may receive a question from another person, read or listen to the question, and understand the information being requested. This involves observation and understanding of information, which are mental acts that can practically be performed by a person.
The limitation of “initiating a session with the first network of agents”, as drafted, describes beginning an inquiry or consultation in response to the received query. A person can mentally begin considering a received question and determine that one or more sources of information should be consulted to answer the question.
The limitation of “a front man agent having access to a first large language model”, as drafted, describes using a primary intermediary having access to a first source of information. A person can act as a primary point of contact, receive a question, and consult a first source of information relevant to that question.
The limitation that “the front man agent has access to at least a first branch agent” and that “the first branch agent access to a second large language model”, as drafted, describes permitting the primary intermediary to consult a second intermediary having access to a different source of information.
For example, a person receiving a question may first consult one source of information and then communicate with another person who has access to a different source. The person may mentally determine that the second source should be consulted based on the nature of the question.
These activities involve receiving information, understanding the information, determining which sources should be consulted, and coordinating access to those sources. Such activities involve observation, evaluation, judgment, and the mental organization of information. They can be practically performed in the human mind, either alone or with the aid of pen and paper.
Accordingly, the identified claim limitations describe the mental process of receiving a question and coordinating consultation with multiple sources of information through primary and secondary intermediaries. Mental processes constitute a recognized grouping of abstract ideas under MPEP § 2106.04(a)(2).
The limitations requiring “a first network of agents,” “at least two different large language models,” “an interface,” and “creating an instance of a front man agent” are treated as additional computer-related elements and are evaluated under Step 2A, Prong Two.
At Step 2A, Prong Two:
The claim recites the additional elements of:
“a first network of agents”; “at least two agents”; “at least two different large language models”; “an interface”; “creating an instance of a front man agent”; “a first large language model”; “a first branch agent”; and “a second large language model.”
However, these additional elements merely amount to adding the words “apply it,” or an equivalent, to the judicial exception, using a computer as a tool to perform the abstract idea, and generally linking the judicial exception to a particular technological environment or field of use. See MPEP §§ 2106.05(f) and 2106.05(h).
The limitation of “an interface for receiving a user’s query” merely uses an interface to obtain the information upon which the mental process is performed. Receiving the query is insignificant pre-solution activity in the form of data gathering because the query must first be obtained before it can be processed or directed to different sources. See MPEP § 2106.05(g).
The limitation of “initiating a session with the first network of agents” merely instructs generally recited software components to begin performing the abstract information-access process. The claim does not recite a particular communication protocol, networking technique, or technical procedure for initiating the session.
The limitation that “at least two agents facilitate access to at least two different large language models” merely uses software agents and large language models as computerized tools for consulting different sources of information. The claim states the desired result of facilitating access but does not recite how such access is technically established, controlled, secured, or maintained.
The limitation that initiating the session includes “creating an instance of a front man agent” merely initializes the software intermediary used to perform the abstract idea. The claim does not recite a particular process for creating the instance, allocating computing resources to the instance, configuring the instance, or improving the operation of the computer or agent.
The limitation that the front man agent has access to a first large language model and has access to a branch agent associated with a second large language model merely assigns the abstract information-consultation functions to computerized components.
The claim does not recite:
how the user’s query is analyzed or divided;
how a large language model is selected;
how the agents communicate;
how prompts are generated for the different large language models;
how responses from the models are evaluated or combined;
how conflicting responses are resolved;
how computing or network resources are reduced; or
how the operation of an agent, large language model, network, or computer is improved.
Rather, the claim merely recites the desired result of providing single-point access to different large language models through a hierarchy of software agents.
The preamble language “for facilitating single point of user access to multiple large language models to generate a response to a user query” merely identifies an intended use and desired result. It does not require a specific technological mechanism for generating the response or improving the functioning of the large language models.
Viewing the additional limitations together and the claim as a whole, the claim merely uses an interface, network, software agents, and large language models as tools to automate the mental process of receiving a question and consulting multiple sources of information. Nothing in the claim integrates the judicial exception into a practical application.
Therefore, claim 1 is directed to an abstract idea under Step 2A.
At Step 2B:
The conclusions regarding the mere implementation of the abstract idea using computer components are carried over and do not provide significantly more than the judicial exception.
The limitation of “an interface for receiving a user’s query” performs the generic computer function of receiving data. Receiving or transmitting data over a network is a well-understood, routine, and conventional computer activity recognized in MPEP § 2106.05(d)(II), citing, for example, Intellectual Ventures I LLC v. Symantec Corp., 838 F.3d 1307, 1321, 120 USPQ2d 1353, 1362 (Fed. Cir. 2016), and buySAFE, Inc. v. Google, Inc., 765 F.3d 1350, 1355, 112 USPQ2d 1093, 1096 (Fed. Cir. 2014).
The “first network of agents” merely provides communication and access between software components. The “front man agent” and “first branch agent” act as computerized intermediaries, while the first and second large language models act as computerized information sources.
The limitation of “creating an instance of a front man agent” merely initializes the software component used to perform the abstract information-routing process. No particular agent-instantiation technique or unconventional computing architecture is claimed.
As an ordered combination, the claim provides an interface that receives a query, creates a primary software intermediary, and permits that intermediary to access a first information source and another intermediary associated with a second information source. This arrangement merely implements the abstract process through generally recited computer components performing their ordinary information-receiving, communication, initialization, and information-access functions.
Accordingly, at Step 2B, these additional elements, both individually and in combination, do not amount to significantly more than the judicial exception. Therefore, claim 1 is ineligible under 35 U.S.C. § 101.
In regards with claims 2-3 “wherein the front man agent and the first branch agent have access to one or more coded tool nodes and wherein the front man agent processes the user's query against the first large language model and determines that access to the first branch agent and the second large language model is necessary to answer the user's query”, as drafted, is a process that, under its broadest reasonable interpretation covers the process of process of mental processes. The claim has not added any additional elements that could integrate the judicial exception into a practical application or provide significantly more than the abstract idea.
In regards with claims 4-6, 9, “wherein the system includes a dictionary of user data which is designated as private by the user and further wherein, when the user's query includes at least some data which is designated as private, the system applies pre-established rules regarding use of the private data in one or more downstream actions and a registry of individual agents available to the user via the single point of user access, wherein the registry includes at least a list of each individual agent in the first network of agents, and wherein the registry includes a listing of data-only dictionaries of each individual agent specification for each individual agent in the first network, the dictionaries being in the form of a key/value mapping, and wherein the registry of agents further includes one or more dictionaries of the agent specification for one or more external agents accessible to the user via the single point of user access, wherein the one or more external agents are located in a different network of agents from the first network of agents”, as drafted, is a process that, under its broadest reasonable interpretation covers the process of process of mental processes. The claim has not added any additional elements that could integrate the judicial exception into a practical application or provide significantly more than the abstract idea.
In regards with claim 7, “wherein the data-only dictionaries of each agent specification may be read from one or more files selected from the following text-based formats: JSON (JavaScript Object Notation), Hocon (Human-Optimized Configuration Object Notation), YAML (yet another markup language), or XML (Extensible Markup Language)”, as drafted, is a process that, under its broadest reasonable interpretation covers the process of process of mental processes. The claim has not added any additional elements that could integrate the judicial exception into a practical application or provide significantly more than the abstract idea.
In regards with claim 8, “wherein communications between the front man agent and at least a first branch agent is via chat streaming and further wherein chat details are stored in a journal at both the front man agent and the first branch agent”, as drafted, is a process that, under its broadest reasonable interpretation covers the process of process of mental processes. The claim has not added any additional elements that could integrate the judicial exception into a practical application or provide significantly more than the abstract idea.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1-3 are rejected under 35 U.S.C. 103 as being unpatentable over Seibel et al (US 12,111,859 B2) in view of Yang et al (US 2025/0284561 A1).
As per claim 1, Seibel teaches a system for facilitating single point of user access to multiple large language models to generate a response to a user query from a access, the system comprising:
a first network of agents, wherein at least two agents facilitate access to at least two different large language models (see Fig. 5 and Fig. 14, e.g., teaches an Enterprise Generative Artificial Intelligence System including an Orchestrator Module and a plurality of specialized agent modules including retrieval agents, structured data retrieval agents, unstructured data retrieval agents, API agents, math agents, visualization agents, code generation agents, and additional specialized agents and the orchestrator comprises one or more first large language models, while the selected agents comprise one or more respective second large language models, thereby teaching multiple agents associated with different large language models); and an interface for receiving a user's query (see Fig. 14, steps 1402 and 1404, e.g., teaches obtaining a user input and providing the input to the orchestrator for processing by the enterprise generative artificial intelligence system), initiating a session with the first network of agents (see Fig. 14, steps 1406-1416, e.g., teaches that after receiving the user input, the orchestrator selects one or more agents, routes the input to the selected agents, receives outputs from the selected agents, and generates a response based upon the returned outputs),
Seibel does not explicitly teach wherein initiating the session includes creating an instance of a front man agent having access to a first large language model and further wherein the front man agent has access to at least a first branch agent, the first branch agent having access to a second large language model.
However, Yang teaches wherein initiating the session includes creating an instance of a front man agent having access to a first large language model and further wherein the front man agent has access to at least a first branch agent, the first branch agent having access to a second large language model (see Figs. 3 and 5 and it description, e.g., discloses wherein a multi-agent collaboration-based request processing system including a host agent and an external agent, wherein the agents communicate through a communication channel between agents (140) and it further teaches that the host agent is connected to a first fine-tuned LLM service system (300a) while the external agent is connected to a second fine-tuned LLM service system (300b) and see Fig. 14 teaches receiving a user query, transferring the request to the host agent, determining whether the host agent can complete the requested task, requesting an external agent to perform at least a portion of the requested task when the host agent cannot complete the task, receiving the external processing result from the external agent, and generating a final processing result based upon the returned result).
Thus, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to apply the teachings of Yang with the teachings of Seibel in order to improve the efficiency and flexibility of processing user queries by enabling coordinated task delegation among multiple agents utilizing different large language models, thereby allowing specialized agents to process portions of a user request and return results for response generation (Yang).
As per claim 2, wherein the front man agent and the first branch agent have access to one or more coded tool nodes (see Fig. 5 and Fig. 14, e.g., teaches n enterprise generative artificial intelligence system including an Orchestrator Module configured to coordinate a plurality of specialized software agent modules, the orchestrator invokes and communicates with these specialized software modules to perform respective processing functions associated with a received user request. Likewise, the selected specialized agents utilize their respective software modules to execute the delegated tasks and return results to the orchestrator, these specialized executable software modules correspond to the claimed coded tool nodes, and both the front-man agent (orchestrator) and the branch agents have access to such coded tool nodes for performing their assigned processing operation, Seibel).
As per claim 3, wherein the front man agent processes the user's query against the first large language model (see Fig. 14, teaches an orchestrator that receives a user input and processes the input using one or more first large language models before selecting one or more specialized agents for additional processing, Seibel) and determines that access to the first branch agent and the second large language model is necessary to answer the user's query (See Fig. 5, Fig. 14, and claims 16-18, e.g., teaches a host agent that receives a user request and determines whether it can complete the requested task. In response to determining that additional processing is required, the host agent requests an external agent to perform at least a portion of the requested task. The external agent processes the delegated task using a second fine-tuned large language model service system and returns the processing result to the host agent, which then generates the final response, Yang).
Claim 4 is rejected under 35 U.S.C. 103 as being unpatentable over Seibel et al (US 12,111,859 B2) in view of Yang et al (US 2025/0284561 A1) further in view of Ustyuzhanin et al (US 2025/0317297 A1, claiming priority of provisional application No: 63/575099, filed on Apr.5, 2024).
As per claim 4, Seibel and Yang do not explicitly teach wherein the system includes a dictionary of user data which is designated as private by the user and further wherein, when the user's query includes at least some data which is designated as private, the system applies pre-established rules regarding use of the private data in one or more downstream actions.
However, Ustyuzhanin teaches wherein the system includes a dictionary of user data which is designated as private by the user and further wherein, when the user's query includes at least some data which is designated as private, the system applies pre-established rules regarding use of the private data in one or more downstream actions ([0125], e.g., teaches that "tokenizer 510 stores the secret tokens in a token database (e.g., dictionary 522) as token-data pairs" and that the secret tokens are encrypted using an encryption scheme and paragraph [0127] further teaches "creating a dictionary of secret tokens on the client side, and assigning secret tokens to sensitive information" before processing by the large language model. Thus, the dictionary stores user-designated private information, while the predefined tokenization process constitutes pre-established rules governing the use of the private data during LLM processing).
Thus, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to apply the teachings of the Ustyuzhanin with the teachings of Seibel and Yang in order to improve protection of confidential user information by creating a dictionary of private user data and applying predefined tokenization rules before processing user prompts with a large language model, thereby reducing exposure of sensitive information during LLM processing(Ustyuzhanin).
Claims 5-6 and 9 are rejected under 35 U.S.C. 103 as being unpatentable over Seibel et al (US 12,111,859 B2) in view of Yang et al (US 2025/0284561 A1) further in view of Thompson, III et al (US 2025/0371317 A1).
As per claim 5, Seibel and Yang do not explicitly teach further comprising a registry of individual agents available to the user via the single point of user access, wherein the registry includes at least a list of each individual agent in the first network of agents.
However, Thompson teaches further comprising a registry of individual agents available to the user via the single point of user access, wherein the registry includes at least a list of each individual agent in the first network of agents (see paragraph [0118], e.g., teaches that the automated agent loads information into data registries 224, including registries storing model identifiers, micro-prompt identifiers, skill identifiers, planner identifiers, and data resource identifiers accessible to the agent. The registries are organized according to data type and are dynamically populated for each agent instance and paragraph [0119] teaches that the automated agent includes agent definition 214 and data registries 224, where the agent definition includes information used to initialize the agent instance).
Thus, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to apply the teachings of the Thompson with the teachings of Seibel and Yang in order to organize agent specifications within multiple structured registries, thereby improving storage, discovery, configuration, and management of multiple agents in a distributed multi-agent system (Thompson).
As per claim 6, wherein the registry includes a listing of data-only dictionaries of each individual agent specification for each individual agent in the first network, the dictionaries being in the form of a key/value mapping (see paragraph [0236] teaches that each planner registry is associated with attributes, parameters, inputs, outputs, constraints, AI services, workflows, tools, and relationships, and further teaches that associations among registries may be implemented using keys, pointers, arguments, or other referential mechanisms, Thompson).
As per claim 9, wherein the registry of agents further includes one or more dictionaries of the agent specification for one or more external agents accessible to the user via the single point of user access, wherein the one or more external agents are located in a different network of agents from the first network of agents (see paragraph [0238] teaches that registries and registry entries are specific to different automated agents, and that different users may have different sets of registries depending on user context and preferences and paragraphs [0320]-[0324], [0329]-[0333] teach an agent team including multiple skill-based agents, where agent teams may be dynamically created, agents communicate with one another, and tasks are distributed among specialized agents and a multi-agent marketplace including multiple agent teams that communicate, interact, and exchange information over a network, enabling users to discover and access different automated agents through the application, Thompson).
Claim 7 is rejected under 35 U.S.C. 103 as being unpatentable over Seibel et al (US 12,111,859 B2) in view of Yang et al (US 2025/0284561 A1) in view of Thompson, III et al (US 2025/0371317 A1) further in view of Fedoruk et al (US 12,493,473 B1).
As per claim 7, Seibel, Yang and Thompson do not explicitly teach wherein the data-only dictionaries of each agent specification may be read from one or more files selected from the following text-based formats: JSON (JavaScript Object Notation), Hocon (Human-Optimized Configuration Object Notation), YAML (yet another markup language), or XML (Extensible Markup Language).
However, Fedoruk teaches wherein the data-only dictionaries of each agent specification may be read from one or more files selected from the following text-based formats: JSON (JavaScript Object Notation), Hocon (Human-Optimized Configuration Object Notation), YAML (yet another markup language), or XML (Extensible Markup Language) (see col. 2, lines 8-13, 42-67, e.g., AI Agents integrate various components (Agent Objects) including AI Models, Integrators, Data Sources, Tools, Prompts, Code Blocks, Datasets, Routers, Memory Objects, and others according to defined workflows and rules specified in their 'Manifest Files, The platform user can then select and connect agent objects... generate a manifest file... The manifest file can keep track of specific versions of the agent objects... dependencies... resources... The server can cause the manifest file to be validated…, and see col. 5, lines 29-46 teaches that the manifest files are stored in standardized text-based configuration file formats including YAML and XML and col.29, lines 24-32 teaches one type of storage object can define that output of the AI model and other conversation history is summarized in a particular format (e.g., JavaScript Object Notation (JSON) format, thereby teaching that the agent specification may be read from text-based files).
Thus, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to apply the teachings of the Fedoruk with the teachings of Seibel, Yang and Thompson in order to store and utilize agent specifications in manifest files represented in standardized text-based configuration formats, such as YAML and XML, thereby improving portability, interoperability, configuration management, validation, and deployment of AI agents within a distributed multi-agent system (Fedoruk).
Claims 8 are rejected under 35 U.S.C. 103 as being unpatentable over Seibel et al (US 12,111,859 B2) in view of Yang et al (US 2025/0284561 A1) further in view of Guo et al (US 2026/0111708 A1).
As per claim 8, Seibel and Yang do not explicitly teach wherein communications between the front man agent and at least a first branch agent is via chat streaming and further wherein chat details are stored in a journal at both the front man agent and the first branch agent.
However, Guo teaches wherein communications between the front man agent and at least a first branch agent is via chat streaming and further wherein chat details are stored in a journal at both the front man agent and the first branch agent ([0003], e.g., teaches communications logs between users and LLM agents are analyzed by a reflections agent and generating long-term memory from previous communications to improve future performance of the LLM, while paragraph [0022] teaches analyzing communications, generating text-based samples from the communications, and storing the generated samples in a long-term memory database for later retrieval and use. Figures 3 and 4 further illustrate storing and recalling communications associated with LLM agents through a storage service and reflections protocol).
Thus, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to apply the teachings of the Guo with the teachings of Seibel and Yang in order to improve collaboration between multiple agents by maintaining a history of communications for subsequent retrieval and future task execution (Guo).
It is noted that any citation [[s]] to specific, pages, columns, lines, or figures in the prior art references and any interpretation of the references should not be considered to be limiting in any wav. A reference is relevant for all it contains and may be relied upon for all that it would have reasonably suggested to one having ordinary skill in the art. [[See, MPEP 2123]].
Pertinent Prior Art
The prior art made of record and not relied upon is considered pertinent to
applicant's disclosure.
Bull et al US 2026/0073257 A1 discloses Method for Distributed Artificial Intelligence Agent Communication, Involves Selecting Second Artificial Intelligence Agent to Perform Portion of Task and Sending Request to Second AI Agent By First AI Agent.
Ma et al US 2025/0390295 A1 discloses Method for Performing Software Development Process with Multiple Large Language Model Agents, Involves Determining Estimates for Completion of Respective Work Items of Software Development Process Based on Interactions, And Adjusting Characteristics of Process.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Mohammad A Sana whose telephone number is (571)270-1753. The examiner can normally be reached Monday-Friday 9-5.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Sanjiv Shah can be reached at 5712724098. 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.
/Mohammad A Sana/Primary Examiner, Art Unit 2166
Prosecution Insights