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
The amendment filed on June 09, 2026 has been entered. Claims 1, 12, 14, 19 have been amended. Claims 1-20 are currently pending in the application.
With regard to the claim objection directed towards claim 12, the objection is hereby withdrawn in view of the amendment to those claims.
Response to Arguments
35 U.S.C 103
Applicant’s arguments filed with respect to the rejection(s) of claims 1-20 under U.S.C 103 have been fully considered and are persuasive. Therefore, the rejection has been withdrawn. However upon further consideration and in light of Applicant’s amendments, new grounds of rejection are made in view of Raheja (U.S Pub # 20250258818).
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, 4, 6-10, 13-14, 16, 18-19 are rejected under 35 U.S.C. 103 as being unpatentable over Cartwright (U.S Pub # 20250342826) in view of Pathak (U.S Pat # 12487803) and in further view of Raheja (U.S Pub # 20250258818).
With regards to claim 1, Cartwright discloses a computer-implemented method for dynamically generating output through artificial intelligence (AI) driven data retrieval, processing, and contextual analysis of interconnected data sources, the computer-implemented method performed by one or more processors of a computing system in communication with one or more data sources, the computer-implemented method comprising:
receiving, by the one or more processors, a request from a user device, wherein the request includes a specified task or a query ([0014] customer inquiry);
utilizing, by the one or more processors, a large language model (LLM) to process the request to determine a request purpose ([0014] detect intent using a LLM);
transmitting, by the one or more processors, data corresponding to the request purpose to a retrieval augmented generation (RAG) model ([0055] integrated with a retrieval-augmented generation (RAG) model);
receiving, by the one or more processors, relevant data associated with the request purpose from the RAG model, wherein the RAG model communicates with one or more agent- based Al systems to identify the relevant data from (i) the LLM, (ii) one or more small language models (SLMs), (iii) one or more dynamic data streams associated with one or more interconnected systems, or (iv) one or more generative software systems ([0055] RAG enhances AI agent’s responses with information retrieved from various sources).
Cartwright does not disclose however Pathak discloses:
transmitting, by the one or more processors, the relevant data to a LLM data deconfliction model ([Col. 18-19 lines 35-03] deconfliction large language model);
utilizing, by the one or more processors, the LLM data deconfliction model to processes the relevant data to reconcile, filter, or resolve conflicting or redundant information of the relevant data ([Col. 18-19 lines 35-03] apply deconfliction logic and may attempt to map the keys between the data and data expected in the contract);
generating, by the one or more processors, the output based upon the relevant data, wherein the output is dynamically generated using the one or more generative software systems ([Col. 18-19 lines 35-03] cause the client device to generate modified contextual data that satisfies a declarative expectation); and
outputting, by the one or more processors, the output to a user interface of the user device ([Col. 19 lines 4-26] output to the client device, a notification of the result of a comparison of the contextual data with a declarative expectation).
It would have been obvious for one of ordinary skill in the art before the date the current invention was effectively filed to have Cartwright by Pathak further reconciliate data retrieved from data sources.
One of ordinary skill in the art would have been motivated to make this modification in order to generate contextual information and output the instructions to execute the software application (Pathak [Col. 2 lines 45-60]).
Cartwright does not disclose however Raheja discloses:
determining, by the one or more processors utilizing the LLM data deconfliction model, whether the conflicting or redundant information is similar to a previously resolved conflict stored in a repository of one or more resolved conflicts ([0015] determination that a particular question and answer pair in the cache of question and answer pairs meets a threshold similarity level); and
responsive to the determining that the conflicting or redundant information is similar to the previously resolved conflict, applying, by the one or more processors utilizing the LLM data deconfliction model, a previously validated resolution from the repository to resolve the conflicting or redundant information ([0015] a stored answer to the received query responsive to a determination that a particular question and answer pair in the cache of question and answer pairs meets a threshold similarity level for the information contained in the set of embeddings); or
responsive to the determining that the conflicting or redundant information is not similar to the previously resolved conflict, resolving the conflicting or redundant information and storing, by the one or more processors utilizing the LLM data deconfliction model, the resolution as a learned rule in the repository ([0015] a large language model operation to generate an answer to the query responsive to a determination that no question and answer pair in the cache of question and answer pairs meets the threshold similarity level for the information contained in the set of embeddings).
It would have been obvious for one of ordinary skill in the art before the date the current invention was effectively filed to have Cartwright and Pathak by Raheja to reuse previously stored answers.
One of ordinary skill in the art would have been motivated to make this modification in order to mitigate LLMs computation costs (Raheja [0038]).
Claims 14 and 19 correspond to claim 1 and are rejected accordingly.
With regards to claim 4, Cartwright further discloses:
wherein the RAG model communicates with the LLM to retrieve general-purpose knowledge by analyzing historical training data to identify the relevant data, and wherein the one or more agent-based Al systems facilitate the communication by automating the retrieval process based upon a request context ([0023] database stores transcripts of successful interactions, which the system can retrieve to enhance response quality over time. This not only contributes to the continuous learning and improvement of the AI agent but also optimizes memory usage by ensuring that only relevant and useful data is stored and utilized for model fine-tuning).
Claim 16 corresponds to claim 4 and is rejected accordingly.
With regards to claim 6, Cartwright further discloses:
wherein the one or more dynamic data streams associated with the one or more interconnected systems includes (i) telemetry data from one or more software-defined vehicles, (ii) sensor data from one or more smart houses, (iii) personalized user data from one or more smart devices, or (iv) multi-dimensional data stream from one or more interconnected databases, and wherein the one or more agent-based Al systems facilitate the communication between the RAG model and the one or more interconnected systems by automating a selection of relevant dynamic data based upon a request context ([0037] systems include smart speakers including amazon echo or a google chrome).
Claim 18 corresponds to claim 6 and is rejected accordingly.
With regards to claim 7, Cartwright further discloses:
wherein the telemetry data includes an operational status, one or more diagnostics, or one or more environmental conditions, wherein the sensor data includes temperature data, one or more security camera feeds, or energy consumption, wherein the personalized user data includes one or more health metrics, location information, or one or more user activity logs, and wherein the multi-dimensional data stream includes structured data, one or more real-time data feeds, historical data, business intelligence data, or user generated data ([0039] user video or chat logs).
With regards to claim 8, Cartwright further discloses:
wherein the one or more interconnected databases include one or more of (i) a public cloud storage, (ii) a private cloud storage, (iii) a subscription-based proprietary database, or (iv) an open source knowledge graph and ontologies ([0036] cloud databases).
With regards to claim 9, Cartwright further discloses:
wherein the RAG model communicates with a live web connection to perform real-time web scraping, indexing, or querying of the one or more interconnected databases to supplement the relevant data ([0055] RAG module integrates with the LM to enhance the Al agent's responses by retrieving relevant information from a knowledge base, a memory vector database storing successful interaction transcripts, and external APIs for dynamic information retrieval).
With regards to claim 10, Cartwright further discloses:
wherein the one or more generative software systems include (i) a code generation tool to generate one or more executable programming scripts, (ii) a text generation tool to generate one or more human-readable textual responses, (iii) an image generation tool to create a visual representation based upon the relevant data, (iv) a video generation tool to produce one or more animated or real-world video outputs, (v) an audio generation tool to generate one or more aural responses, or (vi) a media generation tool to develop one or more websites, one or more applications, or one or more interactive environments, and wherein the one or more agent-based Al systems select and configure at least one generative software system based upon the relevant data ([0102] generate responses to the inquiries based on the detected intent using text-to-speech services).
With regards to claim 13, Cartwright further discloses:
wherein the data corresponding to the request purpose includes one or more of (i) metadata associated with the request, (ii) historical user interaction data relevant to the request, (iii) data identifying domain or category of the request, or (iv) contextual parameters associated with the request ([0068] interpreting the user’s input to understand the intent and context).
Claims 2-3, 15, 20 are rejected under 35 U.S.C. 103 as being unpatentable over Cartwright (U.S Pub # 20250342826) in view of Pathak (U.S Pat # 12487803) and in further view of Raheja (U.S Pub # 20250258818) and Fu (U.S Pub # 20250190604).
With regards to claim 2, Cartwright does not disclose however Fu discloses:
processing, by the one or more processors, via the LLM, the request to extract one or more keywords or one or more phrases indicative of a user intent; and
applying, by the one or more processors, via the LLM, one or more natural language processing (NLP) techniques to classify the one or more keywords or the one or more phrases into one or more categories ([0012] process user queries with a LLM using natural language responses).
It would have been obvious for one of ordinary skill in the art before the date the current invention was effectively filed to have Cartwright, Pathak and Raheja by Fu to classify text into categories.
One of ordinary skill in the art would have been motivated to make this modification in order to include an intent classifier (Fu [0015]).
With regards to claim 3, Cartwright does not disclose however Fu discloses:
wherein the one or more categories include one or more general knowledge topics, one or more specialized domains, or one or more task-specific intents ([0023] can classify general questions or specific questions).
It would have been obvious for one of ordinary skill in the art before the date the current invention was effectively filed to have Cartwright, Pathak and Raheja by Fu to classify text into categories.
One of ordinary skill in the art would have been motivated to make this modification in order to include an intent classifier (Fu [0015]).
Claims 15 and 20 correspond to claim 3 and are rejected accordingly.
Claims 5 and 17 are rejected under 35 U.S.C. 103 as being unpatentable over Cartwright (U.S Pub # 20250342826) in view of Pathak (U.S Pat # 12487803) and in further view of Raheja (U.S Pub # 20250258818) and Barton (U.S Pub # 20260044673).
With regards to claim 5, Cartwright does not disclose however Barton discloses:
wherein the RAG model communicates with the one or more SLMs trained upon one or more datasets corresponding to one or more specific domains to retrieve domain-specific knowledge corresponding to the request, and wherein the one or more agent-based Al systems facilitate the communication by automating a selection of a SLM based upon a request context ([0025] creating fine-tuned, small language model (SLMs) that are optimized for the roles or device types of the network devices to which they are deployed [0080] may select a language model from a set of language models based at least in part on the language model being pre-trained for the device type. The language model may be an LLM, an SLM, or another language model. [0084] configuring the language model to receive the device information from the vector database using retrieval-augmented generation (RAG)).
It would have been obvious for one of ordinary skill in the art before the date the current invention was effectively filed to have Cartwright, Pathak and Raheja by Barton to use fine-tuned language models to retrieve specific data domains.
One of ordinary skill in the art would have been motivated to make this modification in order to obtain embeddings from a vector database to provide a small language model (SLM) that responds to prompts from network administrators (Barton [0008]).
Claim 17 corresponds to claim 5 and is rejected accordingly.
Claim 11 is rejected under 35 U.S.C. 103 as being unpatentable over Cartwright (U.S Pub # 20250342826) in view of Pathak (U.S Pat # 12487803) and in further view of Raheja (U.S Pub # 20250258818) and Avusingi (U.S Pub # 20250252270).
With regards to claim 11, Cartwright does not disclose however Avusingi discloses:
assigning, by the one or more processors, via the LLM data deconfliction model, one or more confidence scores to one or more data elements associated with the relevant data based upon (i) a data source, (ii) recency and relevance of the data to the purpose of the request, or (iii)consistency of the data with corroborating information retrieved from the one or more data sources ([0073] determine confidence scores association with an answer to an inquiry based on an evaluation of one or more LLM sources);
ranking, by the one or more processors, via the LLM data deconfliction model, the one or more data elements based upon the one or more confidence scores ([0074] allow for ranking and comparison); and
filtering, by the one or more processors, via the LLM data deconfliction model, the one or more data elements with confidence scores to remove at least one of the one or more data elements with a confidence score below a predetermined threshold ([0073] confidence scores that satisfy a quality threshold).
It would have been obvious for one of ordinary skill in the art before the date the current invention was effectively filed to have Cartwright, Pathak and Raheja by Avusingi to evaluate how trustworthy a data worthy is in determining an answer.
One of ordinary skill in the art would have been motivated to make this modification in order to determine confidence scores for automatically generated text output of Large Language Models (LLMs) based on the source materials used by the LLMs to generate the output (Avusingi [0004]).
Claim 12 is rejected under 35 U.S.C. 103 as being unpatentable over Cartwright (U.S Pub # 20250342826) in view of Pathak (U.S Pat # 12487803) and in further view of Raheja (U.S Pub # 20250258818) and Oztunc (U.S Pub # 20260057089).
With regards to claim 12, Cartwright does not disclose however Raheja discloses:
identifying, by the one or more processors, via the LLM data deconfliction model, one or more semantically equivalent data points through a similarity scoring process, wherein a natural language processing (NLP) technique analyzes a contextual meaning of each data point and assigns a similarity score based upon a semantic proximity of one or more data points, and wherein the one or more data points with similarity scores above a predetermined similarity threshold are grouped together and considered equivalent for the purpose of the request ([0076] a hierarchical frequently asked questions page by clustering question and answer pairs in the cache of question and answer pairs that meet the threshold similarity level).
It would have been obvious for one of ordinary skill in the art before the date the current invention was effectively filed to have Cartwright and Pathak by Raheja to group similar data points together.
One of ordinary skill in the art would have been motivated to make this modification in order to mitigate LLMs computation costs (Raheja [0038]).
Cartwright does not disclose however Oztunc discloses:
filtering, by the one or more processors, via the LLM data deconfliction model, duplicate or overlapping data based upon the similarity score ([0091] deduplication performed on removing identified matching content within a document based on having significant similarity).
It would have been obvious for one of ordinary skill in the art before the date the current invention was effectively filed to have Cartwright, Pathak and Raheja by Oztunc to identify and filter out semantically similar enough words.
One of ordinary skill in the art would have been motivated to make this modification in order to identify a first set of unique content instances based on deduplication (Oztunc [0010]).
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). 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 TONY WU whose telephone number is (571)272-2033. 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 (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.
/TONY WU/ Primary Examiner, Art Unit 2166
Prosecution Insights