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 .
Response to Arguments
Applicant's arguments filed 27 May 2026 have been fully considered but they are not persuasive.
Applicant contends that “Shen . . . does not involve the LLM generating a tailored message directed to an individual service agent based on that agent’s service description,” because in Shen, “the LLM merely assigns tasks to pre-trained expert models by matching task types to model function descriptions.” Remarks at 14.
The Examiner respectfully disagrees. The Examiner does not conclude that the BRI of the claim is necessarily limited to “generating a tailored message . . . based on [an] agent’s service description;” rather, the claim merely recites “generating . . . a first message to a first service agent . . . based on a service description.” The BRI of this limitation may include distributing a generated message to a service agent based on the service agent’s description.
Therefore, since Shen discloses sending generated messages to a service agents based on the service agent’s description (Shen, Section 3.2, “Following task planning, HuggingGPT proceeds to the task of matching tasks with models, i.e., selecting the most appropriate model for each task in the parsed task list. To this end, we use model descriptions as the language interface to connect each model”), the Examiner finds a person having ordinary skill in the art would have found the recited “generating . . . a first message to a first service agent . . . based on a service description” obvious.
Applicant contends that Talebirad does not “describe the responsive receipt of a first agent response that is then included in a plurality of agent responses used to generate a query response.” Remarks at 15.
The Examiner respectfully disagrees. Talebirad discloses “a plugin could act as a communication board, enabling multi-directional communication between multiple agents” and “Each agent can send messages through the channels that it is connected to, and each message m ∈ Mij, sent from agent Ai to Aj, is represented as a tuple m = (Sm,Am,Dm)” (Section 2.3). The Examiner finds that a person having ordinary skill would have found the recited “responsive to sending the first message to the first service agent, receiving a first agent response from the first service agent, the first agent response included in the plurality of agent responses” obvious in view of Talebirad’s disclosure of a plurality of agents connected via a communication channel to send responses to messages.
Claim Rejections - 35 USC § 112
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
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.
Claims 15-20 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
Claim 15 recites the limitation “the system.” There is insufficient antecedent basis for this limitation in the claim. Claims 16-20 depend on claim 15; therefore, they are indefinite for the same reason.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claim(s) 1, 3, 8, 10, 15, and 17 is/are rejected under 35 U.S.C. 103 as being unpatentable over Shen (Yongliang Shen, Kaitao Song, Xu Tan, Dongsheng Li, Weiming Lu, and Yueting Zhuang. 2023. HuggingGPT: Solving AI tasks with ChatGPT and its friends in HuggingFace. arXiv preprint arXiv:2303.17580 (2023)) and further in view of Talebirad (Talebirad Y, Nadiri A. Multi-agent collaboration: harnessing the power of intelligent LLM agents. 2023. ArXiv:2306.03314).
Regarding claim 1, Shen teaches: A method comprising:
producing, by a first large learning model (LLM), a processing plan based on a first prompt, wherein the processing plan comprises a plurality of tasks corresponding to a plurality of services (Section 3.1, “1) An LLM (e.g., ChatGPT) first parses the user request, decomposes it into multiple tasks, and plans the task order and dependency based on its knowledge”);
sending, by a processing device, a plurality of messages corresponding to the plurality of tasks to a plurality of service agents (Section 3.2, “After parsing the list of tasks, HuggingGPT next needs to match the tasks and models, i.e., select the appropriate model for each task in the task list”), wherein sending the plurality of messages comprises:
generating, by the first LLM, a first message to a first service agent of the plurality of service agents based on a service description of the first service agent (Shen, Section 3.2, “Following task planning, HuggingGPT proceeds to the task of matching tasks with models, i.e., selecting the most appropriate model for each task in the parsed task list. To this end, we use model descriptions as the language interface to connect each model”), and
generating a query response based on the plurality of agent responses (Section 3.3, “Once a task is assigned to a specific model, the next step is to execute the task, i.e., to perform model inference. For speedup and computational stability, HuggingGPT runs these models on hybrid inference endpoints. By taking the task arguments as inputs, the models compute the inference results and then send them back to the large language model”).
Shen does not teach; however, Talebirad discloses: the plurality of service agents correspond to the plurality of services and comprise a plurality of second LLMs that produce a plurality of agent responses (Section 2.1, “Each agent i ∈ V is represented as a tuple Ai = (Li,Ri,Si,Ci,Hi), where: Li refers to the language model instance utilized by the agent. This encompasses the model’s type (such as GPT-4 or GPT-3.5-turbo)”); and
responsive to sending the first message to the first service agent (Section 2.3, “[e]ach agent can send messages through the channels that it is connected to, and each message m ∈ Mij, sent from agent Ai to Aj”), receiving a first agent response from the first service agent (Section 2.1, “an agent’s responsibilities could include processing and responding to user queries, coordinating interactions between other agents, managing resources, or overseeing a particular aspect of the system’s operations”), the first agent response included in the plurality of agent responses (Section 2.3, “Each agent can send messages through the channels that it is connected to, and each message m ∈ Mij, sent from agent Ai to Aj, is represented as a tuple m = (Sm,Am,Dm)” and “a plugin could act as a communication board, enabling multi-directional communication between multiple agents”).
It would have been obvious to a person having ordinary skill in the art, at the effective filing date of the invention, to have applied the known technique of the plurality of service agents correspond to the plurality of services and comprise a plurality of second LLMs that produce a plurality of agent responses; and responsive to sending the first message to the first service agent, receiving a first agent response from the first service agent, the first agent response included in the plurality of agent responses, as taught by Talebirad, in the same way to the plurality of service agents, as taught by Shen. Both inventions are in the field of multi-agent LLM systems for complex task solving, and combining them would have predictably resulted in “a collaborative environment where multiple intelligent agent components, each with distinctive attributes and roles, work together to handle complex tasks more efficiently and effectively,” as indicated by Talebirad (abstract).
Regarding claim 3, Talebirad teaches: The method of claim 1, further comprising: generating, by the plurality of second LLMs executing on the plurality of service agents, a plurality of API calls (Section 4.3.1, “In this system, API calls and their documentation are used to instruct the LLM about the specific tasks each API can handle. The model learns to map prompts to API calls by using a retrieval system to access the most up-to-date API documentation from the database”); executing, by the plurality of service agents, the plurality of API calls to their corresponding one of the plurality of services to produce a plurality of API responses (Section 3.1, “By connecting agents to plugins, agents gain access to tools, resources, or external services that enhance their capabilities. These connections allow agents to leverage the functionalities of the plugins”); receiving the plurality of API responses from the plurality of service agents (Section 4.1.1, “Messages sent through these connections may include task assignments, requests for information, or commands to execute certain operations”); and generating the query response based on the plurality of API responses (Section 1, “Our proposed abstraction allows users to engage with a “black box” by providing an initial prompt and receiving the final output without grappling with the underlying complexities of agent collaborations and interactions”).
Claim(s) 8, 10, 15, and 17 recite(s) commensurate subject matter as claim(s) 1 and 3. Therefore, it/they is/are rejected for the same reasons.
Claim(s) 2, 9, and 16 is/are rejected under 35 U.S.C. 103 as being unpatentable over Shen and Talebirad, as applied above, and further in view of Singh (US 2024/0095077).
Regarding claim 2, Shen and Talebirad do not teach; however, Singh discloses: the plurality of agent responses comprise a plurality of Application Programming Interface (API) calls (¶ 108, “the task(s) T1-T4 include action primitives “walk,” “find,” “grab,” and “putin,” respectively, which may each be an API call”), the method further comprising: organizing the plurality of API calls into an execution stack (¶ 108, “the plan 220 includes task(s) T1-T4 that the plan generator functionality 122 generated based at least in part on the prompt 200 of FIG. 2A”); and executing the plurality of API calls to the plurality of services in an order based on the execution stack (¶ 104, “The task(s) (e.g., a task 215) may include one or more Application Programming Interface (“API”) calls to action primitives (e.g., “grab(‘wineglass’)”)” and ¶ 105, “The comments (e.g., a comment 216) may provide natural language summaries for subsequent sequences of actions”).
It would have been obvious to a person having ordinary skill in the art, at the effective filing date of the invention, to have applied the known technique of the plurality of agent responses comprise a plurality of Application Programming Interface (API) calls, the method further comprising: organizing the plurality of API calls into an execution stack; and executing the plurality of API calls to the plurality of services in an order based on the execution stack, as taught by Singh, in the same way to the plurality of agent responses, as taught by Shen and Talebirad. Both inventions are in the field of LLM multi-agent orchestration systems, and combining them would have predictably resulted in a method that “generate(s) a plan to perform a task (identified in the prompt) that is to be performed by an agent (real world or virtual),” as indicated by Singh (¶ 2).
Claims 9 and 16 recite commensurate subject matter as claim 2. Therefore, they are rejected for the same reasons.
Allowable Subject Matter
Claims 4-7, 11-14, and 18-20 are objected to as being dependent upon a rejected base claim, but would be allowable if rewritten in independent form including all of the limitations of the base claim and any intervening claims.
The following is a statement of reasons for the indication of allowable subject matter: No reference or combination of references were uncovered that teach or at least suggest the limitations of “identifying a first task,” “identifying the first service agent,” “determining that the first agent response completes the first task,” and “marking the first task as complete,” as recited in dependent claim 4 as a whole, and commensurately recited in dependent claims 11 and 18.
Conclusion
THIS ACTION IS MADE FINAL. Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to JACOB D DASCOMB whose telephone number is (571)272-9993. The examiner can normally be reached M-F 9:00-5:00.
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, Pierre Vital can be reached at (571) 272-4215. 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.
/JACOB D DASCOMB/Primary Examiner, Art Unit 2198
Prosecution Insights