Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Claims 1-14 are pending.
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)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention.
Claims 1-7 and 10 are rejected under 35 U.S.C. 102(a)(1) as being anticipated by Anthony et al. (US Pub. 2025/0005224 A1).
Regarding Claim 1, Anthony teaches a computerized method (see Fig.1 and paragraph [0029]), performed by a computing system having one or more hardware computer processors and one or more non-transitory computer readable storage device storing software instructions executable by the computing system to perform the computerized method (see Fig.10 (1001,1002,1003) and paragraphs [0102-0103]), comprising:
by an agent service configured to interact with an LLM to complete a run (see Fig.1 (107,103), Fig.2, paragraph [0029] and paragraph [0042], the application running on the computing system communicating with a large language model to complete one or more tasks):
accessing a state machine comprising a plurality of states, each state of the plurality of states comprising a prompt, one or more tools, and an indication of one or more next states (see Fig.1 (107,109), Fig.5, (503,502), paragraph [0034] and paragraphs [0069-0070], accessing the computing system including a user interface displaying prompts for interacting with a remote large language model, wherein the application running on the computing system engages in a series of prompts or queries to the large language model and receiving responses from the large language model);
executing an initial state of the state machine by at least:
providing an initial state prompt to the LLM comprising natural language text configured to cause the LLM to perform a task (see Fig.3 (303), Fig.5 (503) and paragraph [0055], the initial prompt to obtain information to configure or build a system), the initial state prompt further comprising information relating to accessing the state machine and information relating to accessing a dataset to complete the run (see Fig.1 (107), Fig.3 (303), paragraph [0037] and paragraph [0057], access stored user information and information relating saved projects or workflows);
causing a data processing service to implement an initial state tool to retrieve the dataset based on an LLM tool call generated by the LLM responsive to the initial state prompt (see Fig.3 (305) and paragraph [0057], transmitting the prompt to the large language model);
determining a subsequent state of the state machine based on at least an initial LLM output generated by the LLM responsive to the LLM accessing the state machine according to decision logic of the initial prompt (see Fig.3 (307,311), Fig.5 (505), paragraph [0057] and paragraph [0070], the computing system application receiving a response from the large language model and generating a second prompt for query refinement after identifying the context or category information from the first prompt);
and storing the dataset or the initial LLM output to an agent storage for retrieval during the run (see Fig.3 (307,311), Fig.5 (505), paragraph [0057] and paragraph [0070], the application on the computing system receives a response from the large language model and uses the information from the response to generate a second prompt);
and executing the subsequent state of the state machine by at least:
providing a subsequent state prompt to the LLM to cause the LLM to perform a subsequent task, the subsequent state prompt based on the subsequent state of the state machine (see Fig.3 (311,313), Fig.5 (505), paragraphs [0060-0062] and paragraph [0070], transmitting the second prompt to the large language model for data retrieval);
and accessing a subsequent LLM output generated by the LLM responsive to the subsequent state prompt (see Fig.3 (315,317), Fig.5 (507), Fig.6 (607), paragraph [0062] and paragraphs [0071-0072], pump application build or configuration information, help information offer or message processing a request).
Regarding Claim 2, Anthony further teaches generating the state machine based on accessing historical data associated with a historical run of the agent service with the LLM (see Fig.3 (309) and paragraph [0059], historical data including inputs made by a user to the data model over time), the historical run comprising a plurality of non-state-based interactions between the agent service and the LLM iteratively performed by the agent service to complete the historical run (see Fig.2 (203,205) and paragraph [0068], accessing past workflows via embedding database).
Regarding Claim 3, Anthony further teaches generating the initial state prompt or the subsequent state prompt based on at least a historical LLM response (see Fig.3 (309,311) and paragraph [0060], second prompt).
Regarding Claim 4, Anthony further teaches generating the initial state prompt or the subsequent state prompt by providing a historical LLM response to the LLM to update a content or a format of the historical LLM response see Fig.5 (503,505) and paragraphs [0069-0070], generating second prompt refining the query to the large language model after validating response from the large language model).
Regarding Claim 5, Anthony further teaches determining a failure of the subsequent state based on comparing the subsequent LLM output with the dataset or the initial LLM output stored in the agent storage (see Fig.2 (207), paragraph [0050] and paragraph [0069], inaccurate categorization or failure to identify model type category from the first prompt).
Regarding Claim 6, Anthony further teaches in response to determining a failure of the subsequent state, generating a notification to a user requesting user input by the agent service (see Fig.2 (207), Fig.5 (503) and paragraph [0069], regenerating the prompt indicating inaccurate categorization or failure to identify model type category, and enabling user to enter new information).
Regarding Claim 7, Anthony further teaches in response to determining a failure of the subsequent state, executing another state of the state machine by the agent service (see Fig.2 (207), Fig.5 (503) and paragraph [0069], regenerating the prompt and enabling user to enter new information or additional information).
Regarding Claim 10, Anthony further teaches in response to user input to modify the state machine, modifying one or more of the prompt, the one or more tools, or the indication of one or more next states (see Fig.1 (107,109), paragraph [0035], paragraph [0066] and paragraph [0074], edits to an existing data model).
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 8-9 are rejected under 35 U.S.C. 103 as being unpatentable over Anthony et al. (US Pub. 2025/0005224 A1) in view of Singh et al. (US Pub. 2024/0143597 A1).
Regarding Claim 8, Anthony teaches the method of Claim 1 but fails to teach displaying a visualization of the state machine to a user, the visualization of the state machine comprising: a plurality of nodes associated with the plurality of states; and directed edges between the plurality of nodes indicating a process flow of the state machine.
Singh, however, teaches displaying a workflow containing a plurality of nodes associated with the plurality of states and edges between the plurality of nodes indicating a process flow (see Fig.3, paragraph [0015] and paragraphs [0055-0056]).
It would have been obvious for one skilled in the art, before the effective filing date of the application, to include to Anthony’s method the step for displaying a visualization of the state machine to a user, the visualization of the state machine comprising: a plurality of nodes associated with the plurality of states; and directed edges between the plurality of nodes indicating a process flow of the state machine. The motivation would be to enable a user to track the progress of completing a task withing a workflow session with the language model.
Regarding Claim 9, Anthony teaches providing user inputs to modify and update the state machine (see Fig.6 (603,605,609) and paragraphs [0079-0080]) but fails to teach updating the visualization of the state machine.
Singh, however, teaches modifying the visual representation of a workflow to reflect an updated state of the workflow (see Fig.3, paragraph [0015] and paragraph [0021]).
It would have been obvious for one skilled in the art, before the effective filing date of the application, to include to Anthony’s method the step for updating the visualization of the state machine. The motivation would be to display the stages of completing a task within the workflow session.
Claims 11-14 are rejected under 35 U.S.C. 103 as being unpatentable over Anthony et al. (US Pub. 2025/0005224 A1) in view of Mancuso et al. (US Pub. 2024/0403086 A1).
Regarding Claim 11, Anthony teaches a computerized method (see Fig.1 and paragraph [0029]), performed by a computing system having one or more hardware computer processors and one or more non-transitory computer readable storage device storing software instructions executable by the computing system to perform the computerized method (see Fig.10 (1001,1002,1003) and paragraphs [0102-0103]), comprising:
providing an agent service with access to resources to solve a problem with an LLM during a run, the resources comprising at least available tools, data processing services, databases, or an initial prompt (see Fig.1 (107,109,123), paragraphs [0129-0130] and paragraph [0034], the computing system includes an application for generating a user interface displaying a prompt to a large language model, interacting with the large language model and at least one database via a network interface);
causing the agent service to perform the run including interacting with the LLM by providing prompts to the LLM and receiving outputs from the LLM (see Fig.1 (101,103), Fig.5 (503,505) and paragraph [0070], the application running on the computing system provides prompts to the large language models, and enables the computing system to receive responses from the large language model);
maintaining a series of user interactions between the agent service and the large language model during the run (see Fig.5 (503,505) and paragraphs [0068-0070], a series of prompts to the large language model and responses from the large language model), the interactions comprising prompts provided to the large language model by the agent (see Fig.5 (503,505) and paragraphs [0068-0070]), the outputs received by the agent from the large language model (see Fig.5 (503,505) and paragraphs [0070-0071], responses to the prompts), and one or more tools implemented during the run (see Fig.5 (503,505) and paragraphs [0068-0070], displaying the prompts and responses from the large language model on the application user interface);
and generating a state machine based on the interactions (see Fig.5 (503,505,507) and paragraphs [0068-0071] a series of prompts or queries and responses), the state machine comprising a plurality of states executable by the agent to perform another run by interacting with the LLM (see Fig.6 (609) and paragraph [0080], a third prompt to the large language model to generate evolved data model).
Anthony fails to teach maintaining a history log of interactions between the agent service and the large language model, including prompts provided to the large language model and outputs received from the large language model.
Mancuso, however, teaches tracking and displaying interactions between a user account and a large language model in a predicted workflow (see Fig.2 (202,204,206) and paragraph [0050]).
It would have been obvious for one skilled in the art, before the effective filing date of the application, to include to Anthony’s method the step for maintaining a history log of interactions between the agent service and the large language model, including prompts provided to the large language model and outputs received from the large language model. The motivation would be to track a series context information for completing a task within a workflow session with the large language model.
Regarding Claim 12, Anthony further teaches generating the state machine by generating at least a prompt, a set of tools, and a set of next states associated with each of the plurality of states (see Fig.1 (101,103)), Fig.5 (503,505) and paragraph [0067], generating set of prompts to the large language model, and receiving and displaying the responses from the large language model).
Regarding Claim 13, Anthony further teaches generating the state machine based on a user modification to one or more of the prompts, the outputs, or the one or more tools (see Fig,5 (503,505) and paragraphs [0067-0070], a series of prompts to the large language model and responses from the large language model), but fails to teach outputting the history log to a user.
Mancuso, however, teaches tracking and displaying interactions between a user account and a large language model in a predicted workflow (see Fig.2 (202,204,206) and paragraph [0050]).
It would have been obvious for one skilled in the art, before the effective filing date of the application, to include to Anthony’s method the step for outputting the history log to a user. The motivation would be to display the progress of completing a task within a workflow session with the large language model.
Regarding Claim 14, Anthony further teaches generating the state machine based on providing one or more of the prompts or the outputs in the history log to the LLM to update a content or a format of the prompts or the outputs (see Fig5 (503,505) and paragraphs [0067-0070], generating initial prompt to the large language model to obtain category information or initial context information and generating a second prompt based on the initial context information).
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to VU B HANG whose telephone number is (571)272-0582.
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, Hai Phan, can be reached at (571)272-6338. 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.
/VU B HANG/Primary Examiner, Art Unit 2654
Prosecution Insights