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 .
Response to Amendment
Applicant’s Amendments, filed June 8, 2026, have been entered. Claims 1, 2, 4-11, 13-20, and 22-27 have been amended, claims 3, 12, and 21 have been canceled, and claims 1, 2, 4-11, 13-20, and 22-27 are currently pending. The 35 U.S.C. 112(b) rejection of claims 9, 18, and 27 is withdrawn based on amendments.
Response to Arguments
Applicant's arguments filed June 8, 2026 have been fully considered but they are not persuasive. Applicant argues that Dotan-Cohen et al. (Pub. No. US 2025/0225008 A1, hereinafter “Dotan”). Applicant argues that Dotan does not disclose an LLM gateway that renders an API response in an action request response in a natural language (Remarks p. 10). In response, examiner respectfully submits that Dotan teaches that based on the API call generator or the LLM, the LLM interface layer or client layer can receive the API response and transform the data into a format that is consumable by the user, such as visual content that includes a graph in Figs. 5A and 5B (Dotan [0079-0080]). Also see Fig. 5B, 512, the “Summary LLM Walkthrough”, which using the broadest reasonable interpretation discloses natural language in numbered items 1-8 (Dotan [0109-0110]).
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.
Claims 1-2, 4-7, 9-11, 13-16, 18-20, 22-25 and 27 are rejected under 35 U.S.C. 103 as being unpatentable over Dotan-Cohen in view of Cheng et al. (Pub. No. US 2024/0303473 A1, hereinafter “Cheng”).
Regarding claim 1, Dotan teaches:
receiving an action request prompt, in a first natural language, from a user, the action request prompt describing a functional action to be performed by the API gateway, the functional action having a plurality of corresponding functional parameters, the functional action and the functional parameters being agnostic about the operational semantics of the API gateway (Dotan – in Fig. 7, 702 includes receiving an input comprising a user query. In block 704, the process includes determining, based on the input, data for responding to the user query. At 706, the process includes determining, for the data, a schema that characterizes an aspect of the data. At block 708, the process includes generating for an LLM, based on the user query and the schema, a first prompt comprising a first indication associated with the user query, a second indication associated with the schema, and an instruction directing the LLM to generate a set of computer instructions [0118]. The user query interpreter is responsible for receiving an input that includes an initial prompt (or user query, i.e. an action request prompt) intended for the LLM and determining information, such as an intent and contextual information, associated with the user prompt. A “prompt” includes one or more of: a request (for example, a question or instruction [for example, “is my team being productive this month,” or “write a poem”] (i.e. first natural language)), a query, target content, and one or more examples (i.e. functional action) [0052]. Also see API call generator 228, which is part of the LLM interface layer [0051] and is generally responsible for generating an API call associated with computer instructions received from the LLM [0077].)
extracting the functional action and the corresponding functional parameters from the action request prompt, by a large language model (LLM) [gateway], coupled with the API gateway (Dotan – the user query interpreter employs computing logic to infer an intent (i.e. extract functional action) associated with an initial prompt. For example, the intent associated with the initial prompt is determined based on contextual information determined by the context extractor of the user query interpreter. Context extractor utilizes the user activity information and the initial prompt to determine contextual information, which can also be explicitly provided by the user. A context includes information about an initial prompt or related applications and operating system features with which the user is interacting [0053-0054]. Also see API call generator 228, which is part of the LLM interface layer [0051] and is generally responsible for generating an API call associated with computer instructions received from the LLM [0077].)
mapping the extracted functional action and the corresponding functional parameters, by the LLM [gateway], to corresponding API queries having corresponding structured data formats (Dotan – the prompt generator generates a prompt for the LLM that includes at least one of a first indication associated with the user query (received from the user device or from the user query interpreter), a second indication associated with the schema, and an instruction directing the LLM to generate a set of computer instructions. Continuing the example above, the prompt generator generates a prompt that includes (1) “work-life balance” as the first indication indicative of the user query, (2) an indication of the new column (or other data structure) labeled as “average hours worked per week for the finance team” as the second indication indicative of an aspect of the schema, and (3) “generate an API call against the productivity application” as the third indication indicative of instruction directing the LLM to generate a set of computer instructions (i.e. API queries). In this manner, the LLM is prompted to generate an API call [0074]. The prompt generator generates a prompt directing the LLM to generate computer instructions for processing the data in accordance with the user query and schema. For example, the computer instructions from the LLM comprise an API call or parameters for generating an API call [0074-0075].)
and sending the API queries to the API gateway, by the LLM [gateway], in the mapped structured data formats (Dotan – the API call generator utilizes API logic to execute an API call associated with an endpoint based on computer instructions from the LLLM [0077].)
receiving an API response from the API gateway, by the LLM [gateway], the API response being in the structured data formats; rendering the API response, by the LLM [gateway], in an action request response in a second natural language; and returning the action request response to the user (Dotan – the API call generator sends requests to specific API endpoints associated with an API included in the computer instructions from the LLM to retrieve information in a structured format (for example, JSON or XML that would be responsive to the initial request). In this example, the API call generated by the API call generator causes the LLM interface layer or the LLM to receive the information in the structured format [0077]. Based on the API call generated by API call generator or the LLM, the LLM interface layer or client layer can receive the API response and transform the data into a format that is consumable by the user [0078-0079]. Also see Fig. 5B, 512, the “Summary LLM Walkthrough”, which using the broadest reasonable interpretation discloses natural language in numbered items 1-8 (Dotan [0109-0110])
Dotan does not appear to teach:
LLM gateway
LLM gateway
LLM gateway
LLM gateway
LLM gateway
However, Cheng teaches:
LLM gateway (Cheng – the service layer may comprise an AI gateway to communicatively connect to any generative AI model (e.g., in-house built models or models from external servers such as commercialized LLMs) [0030].)
LLM gateway (Cheng – the service layer may comprise an AI gateway to communicatively connect to any generative AI model (e.g., in-house built models or models from external servers such as commercialized LLMs) [0030].)
LLM gateway (Cheng – the service layer may comprise an AI gateway to communicatively connect to any generative AI model (e.g., in-house built models or models from external servers such as commercialized LLMs) [0030].)
LLM gateway (Cheng – the service layer may comprise an AI gateway to communicatively connect to any generative AI model (e.g., in-house built models or models from external servers such as commercialized LLMs) [0030].)
LLM gateway (Cheng – the service layer may comprise an AI gateway to communicatively connect to any generative AI model (e.g., in-house built models or models from external servers such as commercialized LLMs) [0030].)
Accordingly, it would have been obvious to a person of ordinary skill in the art at the time the invention was effectively filed, having the teachings of Dotan and Cheng before them, to modify the system of Dotan with the teachings of Cheng, as indicated above. One would have been motivated to make such a modification to provide a customized generative AI stack that supports a full spectrum of domain-adaptive prompts to enable a full spectrum of personalized and adaptive AI applications (Cheng - [0023]).
Claims 10 and 19 correspond to claim 1 and are rejected accordingly.
Regarding claim 2, Dotan teaches:
tokenizing the action request prompt, (Dotan – see [0041], where tokenization of user prompt is discussed.)
and extracting the functional action and the corresponding functional parameters from the tokenized action request prompt by the LLM [gateway] (Dotan – the user query interpreter employs computing logic to infer an intent (i.e. extract functional action) associated with an initial prompt. For example, the intent associated with the initial prompt is determined based on contextual information determined by the context extractor of the user query interpreter. Context extractor utilizes the user activity information and the initial prompt to determine contextual information, which can also be explicitly provided by the user. A context includes information about an initial prompt or related applications and operating system features with which the user is interacting [0053-0054]. Also see API call generator 228, which is part of the LLM interface layer [0051] and is generally responsible for generating an API call associated with computer instructions received from the LLM [0077].)
Dotan does not appear to teach:
LLM gateway
However, Cheng teaches:
LLM gateway (Cheng – the service layer may comprise an AI gateway to communicatively connect to any generative AI model (e.g., in-house built models or models from external servers such as commercialized LLMs) [0030].)
Accordingly, it would have been obvious to a person of ordinary skill in the art at the time the invention was effectively filed, having the teachings of Dotan and Cheng before them, to modify the system of Dotan with the teachings of Cheng, as indicated above. One would have been motivated to make such a modification to provide a customized generative AI stack that supports a full spectrum of domain-adaptive prompts to enable a full spectrum of personalized and adaptive AI applications (Cheng - [0023]).
Claims 11 and 20 correspond to claim 2 and are rejected accordingly.
Regarding claim 4, Dotan teaches:
wherein the action request prompt is received as a plain text input (Dotan - the user query interpreter is responsible for receiving an input that includes an initial prompt (or user query) intended for the LLM and determining information, such as an intent and contextual information, associated with the user prompt. A “prompt” includes one or more of: a request (for example, a question or instruction [for example, “is my team being productive this month,” or “write a poem”]), a query, target content, and one or more examples [0052].)
Claims 13 and 22 correspond to claim 4 and are rejected accordingly.
Regarding claim 5, Dotan teaches:
training the large language model on a body of text describing the API queries and the corresponding structured data formats of the API gateway (Dotan – Fig. 4 is a block diagram of a language model (for example, a BERT model or Generative Pre-Trained Transformer [GPT]-4 model) that uses particular inputs to make particular predictions (for example, answers to questions or generate computer instructions for generating an API call) [0093].)
Claims 14 and 23 correspond to claim 5 and are rejected accordingly.
Regarding claim 6, Dotan teaches:
periodically updating the LLM [gateway] based on changes in the structured data formats of the API gateway (Dotan – the semantic search engine determines a schema for the identified data sets. The schema comprises a textual description, such as a prose, that describes the data, data types, and/or data structure of the data set [0066]. The data type determiner incorporates aspects of the semantic data into the schema to generate an enriched schema. For example, the schema is updated to include a description of the semantic data and any new data types so that the enriched schema describes a set of data that is usable for addressing an intent or task associated with the user query [0072].
Dotan does not appear to teach:
LLM gateway
However, Cheng teaches:
LLM gateway (Cheng – the service layer may comprise an AI gateway to communicatively connect to any generative AI model (e.g., in-house built models or models from external servers such as commercialized LLMs) [0030].)
Accordingly, it would have been obvious to a person of ordinary skill in the art at the time the invention was effectively filed, having the teachings of Dotan and Cheng before them, to modify the system of Dotan with the teachings of Cheng, as indicated above. One would have been motivated to make such a modification to provide a customized generative AI stack that supports a full spectrum of domain-adaptive prompts to enable a full spectrum of personalized and adaptive AI applications (Cheng - [0023]).
Claims 15 and 24 correspond to claim 6 and are rejected accordingly.
Regarding claim 7, Dotan teaches:
documenting the API queries, the corresponding structured data formats, and the changes in the structured data formats (Dotan – the schema comprises a textual description, such as a prose, that describes the data, data types, and/or data structure of the data set [0066]. the data sample (i.e. documenting) refers to an indication indicative of a previous initial prompt and a previous corresponding schema that were used to generate a target API call. The initial prompt includes a previous query. [0088-0089].)
Claims 16 and 25 correspond to claim 7 and are rejected accordingly.
Regarding claim 9, Dotan teaches:
wherein the first language and the second language are same (Dotan – the API call generator sends requests to specific API endpoints associated with an API included in the computer instructions from the LLM to retrieve information in a structured format (for example, JSON or XML that would be responsive to the initial request). In this example, the API call generated by the API call generator causes the LLM interface layer or the LLM to receive the information in the structured format [0077]. Based on the API call generated by API call generator or the LLM, the LLM interface layer or client layer can receive the API response and transform the data into a format that is consumable by the user [0078-0079].)
Claims 18 and 27 correspond to claim 9 and are rejected accordingly.
Claims 8, 17 and 26 are rejected under 35 U.S.C. 103 as being unpatentable over Dotan in view of Cheng further in view of Fedoruk et al. (Patent No. US 12,493,829 B1, hereinafter “Fedoruk”).
Regarding claim 8, Dotan modified by Cheng does not appear to teach:
versioning the large language model, the API queries, the corresponding structured data formats, and the changes in the structured data formats in accordance with predetermined version control rules
However, Fedoruk teaches:
versioning the large language model, the API queries, the corresponding structured data formats, and the changes in the structured data formats in accordance with predetermined version control rules (Fedoruk – in an AI platform, datasets are typically processed and indexed for efficient retrieval, with metadata management and versioning capabilities [Col. 6 lines 50-59]. The AI platform can implement version control and model lineage tracking to ensure the integrity and reproducibility of AI-generated outputs [Col. 17 lines 9-15]. Also see [Col. 24 lines 17-28], where management policies can relate to data processing. The rules can define and enforce permissible behaviors and configurations for data acquisition, processing, and governance. Data source object parameters can define the origins, methods, and frequency of data acquisition, ensuring data quality, security, versioning, and controlled access.)
Accordingly, it would have been obvious to a person of ordinary skill in the art at the time the invention was effectively filed, having the teachings of Dotan and Cheng before them, to modify the system of Dotan with the teachings of Cheng, as indicated above. One would have been motivated to make such a modification to implement version control and model lineage tracking to ensure the integrity and reproducibility of AI-generated outputs (Fedoruk - [Col. 17 lines 9-15]).
Claims 17 and 26 correspond to claim 8 and are rejected accordingly.
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 RANJIT P DORAISWAMY whose telephone number is (571)270-5759. The examiner can normally be reached Monday-Friday 9:00 AM - 5:00 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, Sanjiv Shah can be reached at (571) 272-4098. 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.
/RANJIT P DORAISWAMY/Examiner, Art Unit 2166
/SANJIV SHAH/Supervisory Patent Examiner, Art Unit 2166
Prosecution Insights