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 .
DETAILED ACTION
Claims 1-20 are pending in Instant Application.
Priority
Examiner acknowledges Applicant’s claim to priority benefits of provisional application 63701480 filed 09/30/2024.
Information Disclosure Statement
The information disclosure statement(s) (IDS) submitted on 05/08/2026 is/are in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement(s) is/are being considered if signed and initialed by the Examiner.
Claim Rejections - 35 USC § 112
The following is a quotation of the first paragraph of 35 U.S.C. 112:
The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same and shall set forth the best mode contemplated by the inventor of carrying out his invention.
The following is a quotation of the second paragraph of 35 U.S.C. 112:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
The following is a quotation of the fourth paragraph of 35 U.S.C. 112:
Subject to the [fifth paragraph of 35 U.S.C. 112], a claim in dependent form shall contain a reference to a claim previously set forth and then specify a further limitation of the subject matter claimed. A claim in dependent form shall be construed to incorporate by reference all the limitations of the claim to which it refers.
Claims 1-3, 8-10, 15-17 rejected under 35 U.S.C. 112, second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which applicant regards as the invention.
The term "complexity level of the task is high", “distributable level of a second task is high”, “distributable level of the task is low”, “complexity level of a second task is low” in claim 1-3, 8-10, 15-17 are a relative term which renders the claim indefinite. The terms are not defined by the claim, the specification does not provide a standard for ascertaining the requisite degree, and one of ordinary skill in the art would not be reasonably apprised of the scope of the invention.
Claims 5-7,11-14, 18-20 are rejected for dependency upon rejected base claim 1, 8, 15 respectively above.
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 1-20 are rejected under 35 U.S.C. 103 as being unpatentable over Thompson et al., “hereinafter Thompson” (U.S. Patent Application: 20250368219) in view of Bhat et al., “hereinafter” (U.S. Patent Application: 20260012431).
As per Claim 1, Thompson discloses a computer-implemented method for executing a task using an optimal architecture, comprising:
selecting, by an analytical agent using one or more processors and a large language model (LLM), the optimal architecture associated with the task (Thompson, Para.39, GAI models, and more specifically, large language models (LLMs), have demonstrated the ability to perform relatively simple tasks (e.g., single-step tasks or tasks that do not include any sub-tasks) using a conversational natural language question and answer format. However, using LLMs to build autonomous agents that can perform more complex tasks (e.g., multi-step tasks or tasks that have one or more sub-tasks), Para.165, Actions (e.g., tools or skills) can be performed, for example, by a planner agent explicitly or through an LLM selection. For instance, a tool can be configured to, given some context (e.g., task to be performed), select and invoke the most appropriate action), comprising:
determining a complexity level of the task is high based on identifying a requirement of a collaboration between a plurality of decentralized agents when executing the task (Thompson, Para.158, To generate a plan, an agent can draft a list of tasks in order of priority (e.g., a to-do list). Each task can be tagged with a status, a deadline, and any resource requirements. The agent can periodically review tasks and update tasks based on the task status and any new information. Each task is or includes an action, including a tool or a skill, which can include one more other workflows. For more complex workflows, the agent can formulate a plan with well-defined goals and concrete actions required to achieve them., Para.142, The agent planners 404 and action planners 408 can be associated with different levels of abstraction, such as high-level, intermediate-level, or low-level, depending on the granularity or complexity of the workflows 406 and the plans 410. For example, a high-level agent planner 404 can use a high-level workflow 406 that specifies the overall objective and the main steps or tasks to achieve it, while a low-level agent planner 404 can use a low-level workflow 406 that specifics the detailed functions or actions to perform each step or task);
determining a distributable level of the task is low based on identifying a requirement of a central agent for task allocation when executing the task (Thompson, Para.142, The agent planners 404 and action planners 408 can be associated with different levels of abstraction, such as high-level, intermediate-level, or low-level, depending on the granularity or complexity of the workflows 406 and the plans 410. For example, a high-level agent planner 404 can use a high-level workflow 406 that specifies the overall objective and the main steps or tasks to achieve it, while a low-level agent planner 404 can use a low-level workflow 406 that specifics the detailed functions or actions to perform each step or task, Para.450, lower-level workflows can be configured to execute functions like memory accesses, reads, and writes, while higher-level workflows can be configured to perform other or higher level tasks and the higher-level workflows may call the lower-level workflows to perform their respective lower-level tasks.); and
selecting a hybrid architecture as the optimal architecture associated with the task, wherein the hybrid architecture comprises a centralized architecture comprising the central agent and a decentralized architecture comprising the plurality of decentralized agents (Thompson, Para.53, Agent topology can refer to the arrangement of agents and distribution of work among a plurality of agents in a system, Para.68, the automated agent 102 can be dynamically configured or reconfigured to perform a task or a series of tasks, via one or more components of the distributed multi-agent system 105, Para.71, , the distributed multi-agent system 105 includes a plurality of sub-agents 106A, 106B, . . . , 106N, a communication service 108, an adaptive machine learning service 110, and a multi-layer memory structure 111.);
assigning the task to an agent associated with the hybrid architecture (Thompson, Para.42, The term role may be used herein to refer collectively to a group or category of tasks, actions, and/or capabilities of, assigned to, or associated with an agent.); and
executing the task using the hybrid architecture, wherein the executing comprises assigning, by the central agent, a step of the task to at least one of the plurality of decentralized agents, and wherein the plurality of decentralized agents collaborate with each other for executing the step of the task (Thompson, Para.107, the automated agent 102 can invoke one or more sub-agents of the distributed multi-agent system 105. For example, a planner sub-agent of the distributed multi-agent system 105 can be invoked to generate a plan for responding to the input using as input a workflow, a profile, and a context model obtained from one or more of the memory layers. The plan can include a plurality of actions that need to be performed (e.g., in sequence or in parallel), where each action has an associated action sub-agent. Each action sub-agent can operate in a similar manner as the planner sub-agent. For example, an action sub-agent can generate a sub-plan for performing its assigned action in a similar manner as the planner sub-agent generates the plan. In this way, the distributed multi-agent system 105 can provide hierarchical planning in which any plan can be expanded to include a plurality of sub-plans, where each sub-plan is performed by a sub-agent and the output is returned to the planner agent or the calling sub-agent, as the case may be. At any level of the planning hierarchy, the respective agent 102 or sub-agent 106 can coordinate the generation and execution of the respective plan and any sub-plans.).
However Thompson does not disclose in response to determining the complexity level of the task being high and the distributable level of the task being low.
Bhat discloses determining the complexity level of the task being high and the distributable level of the task being low (Bhat, Para.58, The task management engine 116 generates task dependency graphs that specify the execution order for subtasks, identifying which subtasks must complete before others can begin, and/or which subtasks can execute in parallel to optimize overall completion time. In some implementations, the task management engine 116 uses recursive decomposition to break large tasks into progressively smaller subtasks until each subtask matches the capabilities of available agents, Para.60, The AI orchestration layer 114 includes a route management engine 120 that determines one or more task routing paths through the distributed agent network 124 to minimize latency, reduce costs, and balance computational loads across available agents. The route management engine 120 maintains network topology maps that track communication pathways, bandwidth capacities, and/or latency measurements between different agents and network nodes within the distributed system.).
It would have been obvious for one of ordinary skill in the art before the effective filing date of the claimed invention to utilize the teachings as in Thompson with the teachings as in Bhat. The motivation for doing so would have been for determine a degree of expected utility of candidate actions by evaluating the actions against the agent's objective function and select executable actions that align with the agent's assigned objectives within the imposed operational constraints or boundaries set by the system the agent is interacting with. However, unlike rule-based programming, AI agents employ machine learning algorithms to assess data and determine actions based on probabilistic models. While this grants AI agents autonomy for real-time operation, AI agent operations are less transparent than rule-based programming, which can result in unpredictable or unverified behaviors within the system. (Bhat, Para.06).
With respect to Claim 8, 15 are substantially similar to Claim 1 and are rejected in the same manner, the same art and reasoning applying.
As per Claim 2, Thompson in view of Bhat discloses the computer-implemented method of claim 1, further comprising:
determining a complexity level of a second task is low based on identifying the requirement of the collaboration between the plurality of decentralized agents when executing the second task (Thompson, Para,254, the processing device creates a second workflow, second micro-prompts, and a second context model by applying an adaptive machine learning process during the first execution of the automated agent, Para.142, while a low-level agent planner 404 can use a low-level workflow 406 that specifics the detailed functions or actions to perform each step or task. The agent planners 404 and action planners 408 can communicate and coordinate with each other to align their workflows and plans 410 and resolve any conflicts or dependencies.); and selecting, when the complexity level of the second task is low, the centralized architecture as the optimal architecture associated with the second task, wherein the centralized architecture comprises: a supervising agent configured to orchestrate a plurality of activities, and a resource allocation configured to reduce redundancy associated with the plurality of activities (Thompson, Para,450, lower-level workflows can be configured to execute functions like memory accesses, reads, and writes, while higher-level workflows can be configured to perform other or higher level tasks and the higher-level workflows may call the lower-level workflows to perform their respective lower-level tasks, Para.51, the orchestrator 216 generates and executes a plan that manages the execution of actions, workflows, and adaptive machine learning processes by automated agent 212 to accomplish a goal, perform a task, or achieve an objective, Para.129, the adaptive machine learning-based orchestrator 216 can load a workflow to accomplish the objective given the event and the context model. The workflow can be a sequence or an arrangement of steps, tasks, actions, or functions that the automated agent can perform (alone and/or via delegation to one or more sub-agents) to achieve the objective. The workflow can be predefined, customized, or generated by the adaptive machine learning-based orchestrator 216 based on the event, the context model, and/or the available AI services 230, tools 232, and data resources 226. The workflow can include various parameters, conditions, constraints, dependencies, or rules that govern the execution of the workflow. The workflow can also include various micro-prompts, which are small units of interaction or communication that the automated agent 200 can use to invoke an AI service. ).
With respect to Claim 9, 16 are substantially similar to Claim 2 and are rejected in the same manner, the same art and reasoning applying.
As per Claim 3, Thompson in view of Bhat discloses the computer-implemented method of claim 1, further comprising:
determining a distributable level of a second task is high based on identifying the requirement of the central agent for task allocation when executing the second task; and selecting, when the distributable level of the second task is high, the decentralized architecture as the optimal architecture associated with the second task, wherein the decentralized architecture comprises the collaboration between the plurality of decentralized agents for sharing information and coordinating one or more actions associated with the second task (Thompson, Para,71, the distributed multi-agent system 105 includes a plurality of sub-agents 106A, 106B, . . . , 106N, a communication service 108, an adaptive machine learning service 110, and a multi-layer memory structure 111, Para.72, The sub-agents 106A, 106B, . . . , 106N cooperate and coordinate with each other to perform tasks on behalf of the user. Each sub-agent 106A, 106B, . . . , 106N can have a specific role or function, such as a profile sub-agent, a planner sub-agent, a workflow sub-agent, a memory sub-agent, or any other sub-agent that can assist the automated agent 102 in executing tasks and fulfilling user requests or goals.,Para.255, the processing device executes the automated agent (e.g., the automated agent instance created at operation 752) using the second micro-prompts, the second planner, the second workflow, and the memory, Para.324, The skill-based agents 1102 and/or agents of the agent team 1104 can communicate, interact, or exchange information with each other using a collective memory 1106. The collective memory 1106 can be a shared memory, a distributed memory, a decentralized memory, or any combination of any of the foregoing and/or other memories, which can store, retrieve, and/or update information related to a workflow, plan, task or sub-task.).
With respect to Claim 10, 17 are substantially similar to Claim 3 and are rejected in the same manner, the same art and reasoning applying.
As per Claim 4, Thompson in view of Bhat discloses the computer-implemented method of claim 1, wherein the hybrid architecture comprises an adaptive architecture selection configured to select between the centralized architecture and the decentralized architecture based on a real-time assessment of a task requirement, the complexity level, and a network condition for executing the task (Thompson, Para,142, The architecture 400 can include agent planners 404, which can use workflows 406 and action planners 408 to create, modify, or optimize plans 410. The agent planners 404 and action planners 408 can be associated with different levels of abstraction, such as high-level, intermediate-level, or low-level, depending on the granularity or complexity of the workflows 406 and the plans 410. For example, a high-level agent planner 404 can use a high-level workflow 406 that specifies the overall objective and the main steps or tasks to achieve it, while a low-level agent planner 404 can use a low-level workflow 406 that specifics the detailed functions or actions to perform each step or task, Para,326, The skill-based agent team 1104 and/or agents of the agent team 1104 can operate in a dynamic, flexible, or adaptive manner to perform a workflow, plan, task or sub-task. For example, the skill-based agent team 1104 and/or agents of the agent team 1104 can create, modify, or update the composition, the configuration, the structure, or the topology of the skill-based agents 1102 based on the requirements, the specifications, the constraints, or the preferences of user, workflow, plan, task or sub-task, Para.53, Agent topology can refer to the arrangement of agents and distribution of work among a plurality of agents in a system. One of the technical problems in this area is how to balance the load and the performance of an agent system, especially when agents are distributed across different locations and networks, to improve stability and optimize the use of power and/or other resources.).
With respect to Claim 11, 18 are substantially similar to Claim 4 and are rejected in the same manner, the same art and reasoning applying.
As per Claim 5, Thompson in view of Bhat discloses the computer-implemented method of claim 1, further comprising:
providing access to a tool registry and an action model to the agent, wherein the tool registry comprises a content repository (Thompson, Para,54, decentralized and distributed repositories of information that can be accessed by potentially any agent of the system.), a structured query language (SQL) (Thompson, Para,501, query technologies, e.g., SQL or NoSQL.), a NoSQL (Thompson, Para,78, NOSQL document stores) and graph databases application programming interface (API), a vector store (Thompson, Para,40, vector databases, graph databases, relational databases, and key-value stores.), a search engine (Thompson, Para,499, search engines,), and a real-time data stream (Thompson, Para,501, real-time data processing can be referred to as a real-time data store), and wherein the action model comprises a reinforcement learning or genetic algorithm, a causality technique, and a predefined machine learning model (Thompson, Para,37, A generative artificial intelligence (GAI) model or generative model uses artificial intelligence technology, e.g., machine learning models, e.g., neural networks, to machine-generate digital content based on model inputs and the previously existing data with which the model has been trained, Para.129, the adaptive machine learning-based orchestrator 216 can load a workflow to accomplish the objective given the event and the context model. The workflow can be a sequence or an arrangement of steps, tasks, actions, or functions that the automated agent can perform (alone and/or via delegation to one or more sub-agents) to achieve the objective.).
With respect to Claim 12, 19 are substantially similar to Claim 5 and are rejected in the same manner, the same art and reasoning applying.
As per Claim 6, Thompson in view of Bhat discloses the computer-implemented method of claim 1, further comprising:
providing an adaptive operation mode to the agent, wherein the adaptive operation mode is configured to enable the agent to provide an immediate response and solution to the task (Thompson, Para,53, provide a technical solution by configuring adaptive machine learning processes and/or observer agents to monitor and regulate the operations of the agents, as well as the communication and the coordination among agents, Para.57, provide a technical solution that involves using an adaptive machine learning approach that can dynamically reconfigure agents and workflows in response to context changes. ).
With respect to Claim 13 are substantially similar to Claim 6 and is rejected in the same manner, the same art and reasoning applying.
As per Claim 7, Thompson in view of Bhat discloses the computer-implemented method of claim 1, further comprising:
providing a security and compliance to the agent, wherein the security and compliance is configured to ensure the step of the task complies with an industry standard and regulation associated with the task (Thompson, Para.41, provide multiple levels of data security, for instance to prevent unauthorized access to user-sensitive data while allowing access to non-sensitive data, in a multi-agent environment, Para.85, A context model can include various types of information, such as preferences, policies, profile data, historical user activity data, sensor data, network data, model parameters, and/or any other data that can be used to configure an agent, workflow, plan, or task, Para.458, a policy can specify that a first memory layer is to be accessed before any other layer of the memory structure can be accessed, or that an agent is required to have access to the first layer in order to access one or more other layers, Para.50, configure and use one or more observer agents and/or adaptive machine learning processes to regulate the agent's output proactively before the output is presented to the user to reduce the need for corrective user feedback, Para.493, concepts include topics, industries, and skills).
With respect to Claim 14, 20 are substantially similar to Claim 7 and are rejected in the same manner, the same art and reasoning applying.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to NORMIN ABEDIN whose telephone number is (571)270-5970. The examiner can normally be reached Monday to Friday from 10 am to 6 pm.
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, Vivek Srivastava can be reached at 5712727304. 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.
/NORMIN ABEDIN/Primary Examiner, Art Unit 2449
Prosecution Insights