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 . Claims 1-20 have been reviewed and are under consideration by this office action.
Notice to Applicant
The following is a Final Office action. Applicant, on 03/02/2026, amended claims. Claims 1-20 are pending in this application and have been rejected below.
Response to Amendment
Applicant’s amendments are received and acknowledged.
The amended claims overcome the 102 rejection by adding new limitations to the independent claims. However, a new 103 rejection is facilitated by the amendments.
Response to Arguments - 35 USC § 101
Applicant’s arguments with respect to the 35 USC 101 rejections have been fully considered, but they are not persuasive.
Applicant contends that the claims require an AI agent system, memory, tools, and a planner which are inherently computer-implemented and depend on multiple AI components and as such are not mental processes.
Examiner respectfully disagrees. The cited elements are each additional elements (recited at a high level of generality and as such are performing the steps would be no more than mere instructions to apply the exception using a generic computer component. See MPEP 2106.05(f) and/or amounts to no more than generally linking the use of the judicial exception to a particular technological environment or field of use – see MPEP 2106.05(h). The claims recite the mental processes of providing requests, breaking prompts into subparts, receiving responses to requests, and evaluating performance all of which are concepts capable of being performed in the human mind (i.e. via pen and paper).
Applicant contends that claims cannot be oversimplified and further points to Desjardins asserting that the claims recite concrete AI-agent evaluation framework with specific technical architecture and workflow.
Examiner respectfully disagrees. The claims do not follow the fact pattern of Desjardins as Dejardins is directed towards training a machine learning model on a series of tasks with the credited benefits of reduced storage, reduced system complexity, and preservation of performance attributes associated with earlier tasks during subsequent computational task (i.e. catastrophic forgetting). The present claims recite a system for providing an AI system with requests, breaking the requests into a plurality of sub-parts, formulating a sequence of actions, generating an answer, receiving a response, evaluating the performance of the agent, and storing the metrics.
The 101 Rejection is updated and maintained below.
Response to Arguments - 35 USC § 102/103
Applicant’s amendments have overcome the 102 rejection, but facilitate a new 103 rejection. The remaining arguments are moot in view of the new line of 103 Rejections seen below.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to a judicial exception (i.e., a law of nature, a natural phenomenon, or an abstract idea) without significantly more.
Step One - First, pursuant to step 1 in the January 2019 Guidance on 84 Fed. Reg. 53, the claim(s) is/are directed to statutory categories.
Step 2A, Prong One – The claims are found to recite limitations that set forth the abstract idea(s), namely in independent claims recite a series of steps for the abstract idea recited below.
Regarding independent claim(s), (additional elements bolded)
Regarding Claim(s) 1, 8, and 15, A method comprising steps of: operating, in a test environment prior to deployment, an Artificial Intelligence (AI) agent system that includes an agent core connected to a memory module, one or more tools, and a planner;
providing the Al agent system with one or more requests including one or more legitimate requests and one or more malicious requests associated with techniques for compromising Al agents including prompt injection;
in response to the one or more requests, utilizing the planner to break at least one of the one or more requests into a plurality of sub-parts and to formulate a sequence of actions based on the plurality of sub-parts; and generating, by the agent core, an answer using the plurality of sub-parts with the memory module and the one or more tools
receiving a response to each of the one or more requests;
evaluating performance of the Al agent system based on the responses by determining (a) whether the responses sufficiently match expected responses for the one or more legitimate requests and (b) whether the Al agent system refrains from providing answers to the one or more malicious requests
evaluating performance of the Al agent based on responses to each of the one or more requests.
storing results and performance metrics of the evaluation in a result store.
Further regarding Claim 8, A non-transitory computer-readable storage medium having computer-readable code stored thereon for programming one or more processors to perform steps of:
Further regarding Claim 15, A cloud-based system comprising: one or more processors; and memory storing computer-executable instructions that, when executed, cause the one or more processors to:
As drafted, this is, under its broadest reasonable interpretation, within the Abstract idea groupings of “Mental processes—concepts performed in the human mind” (observation, evaluation, judgment, opinion) as the claims are directed towards providing requests, breaking prompts into subparts, receiving responses to requests, and evaluating performance all of which are concepts capable of being performed in the human mind (i.e. via pen and paper).
Further the claims are directed towards the abstract idea grouping of “Certain methods of organizing human activity” — commercial or legal interactions (including agreements in the form of contracts; legal obligations; advertising, marketing or sales activities or behaviors; business relations) and/or managing personal behavior or relationships or interactions between people (including social activities, teaching, and following rules or instructions) as the claims are directed towards providing insights on which helps understand various aspects of business, customers, and products (See Specification, [102]).
Step 2A, Prong Two - This judicial exception is not integrated into a practical application. The independent claims utilize at least the additional elements bolded above. The additional elements are performing the steps would be no more than mere instructions to apply the exception using a generic computer component. See MPEP 2106.05(f) and/or amounts to no more than generally linking the use of the judicial exception to a particular technological environment or field of use – see MPEP 2106.05(h).
Step 2B - The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above with respect to integration of the abstract idea into a practical application, the additional elements are just “apply it” on a computer. (See MPEP 2106.05(f) – Mere Instructions to Apply an Exception – “Thus, for example, claims that amount to nothing more than an instruction to apply the abstract idea using a generic computer do not render an abstract idea eligible.” Alice Corp., 134 S. Ct. at 235) and/or amounts to no more than generally linking the use of the judicial exception to a particular technological environment or field of use – see MPEP 2106.05(h).
Regarding Claims 4-5, 7, 11-12, 14, 18, and 20, the claim further narrows the abstract idea or recite additional elements previously addressed in the independent claims.
Regarding Claims 2, 9, and 16, the claim further recite the additional element(s) of an AI-assisted amplification process. This element(s) is performing the steps would be no more than mere instructions to apply the exception using a generic computer component. See MPEP 2106.05(f) and/or amounts to no more than generally linking the use of the judicial exception to a particular technological environment or field of use – see MPEP 2106.05(h) in Steps 2A-Prong 2 and 2B.
Regarding Claims 3, 10, and 17, the claim further recite the additional element(s) of Large Language Model. This element(s) is performing the steps would be no more than mere instructions to apply the exception using a generic computer component. See MPEP 2106.05(f) and/or amounts to no more than generally linking the use of the judicial exception to a particular technological environment or field of use – see MPEP 2106.05(h) in Steps 2A-Prong 2 and 2B.
Regarding Claims 6, 13, and 19, the claim further recite the additional element(s) of a processing agent, scanning the web and the external feeds. This element(s) is performing the steps would be no more than mere instructions to apply the exception using a generic computer component. See MPEP 2106.05(f) and/or amounts to no more than generally linking the use of the judicial exception to a particular technological environment or field of use – see MPEP 2106.05(h) in Steps 2A-Prong 2 and 2B.
Accordingly, the claim fails to recite any improvements to another technology or technical field, improvements to the functioning of the computer itself, use of a particular machine, effecting a transformation or reduction of a particular article to a different state or thing, adding unconventional steps that confine the claim to a particular useful application, and/or meaningful limitations beyond generally linking the use of an abstract idea to a particular environment. See 84 Fed. Reg. 55. Viewed individually or as a whole, these additional claim element(s) do not provide meaningful limitation(s) to transform the abstract idea into a patent eligible application of the abstract idea such that the claim(s) amounts to significantly more than the abstract idea itself.
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 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 set forth in Graham v. John Deere Co., 383 U.S. 1, 148 USPQ 459 (1966), that are applied 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, 8, and 15 are rejected under 35 U.S.C. 103 as being unpatentable over Morales et al. "LangBiTe: A Platform for Testing Bias in Large Language Models", ARXIV.ORG, CORNELL UNIVERSITY LIBRARY, 201 OLINE LIBRARY CORNELL UNIVERSITY ITHACA, NY 14853, 29 April 2024 in view of Kotte et al. (US 20250252265 A1). And Kiciman, Emre et al.; “Benchmarking and Defending Against Indirect Prompt Injection Attacks on Large language Models,” Arvix.org, Association for Computing Machinery, 21 Dec 2023.
Regarding Claims 1, 8, and 15, Morales teaches: A method comprising steps of: operating, in a test environment prior…, an Artificial Intelligence (AI) agent system that includes an agent core connected to memory, one or more tools, and a planner; (Morales, [pg.3, para. 2]; The complete testing process is controlled by the facade LangBiTe, which is responsible for orchestrating the stages of test case generation, test execution and reporting. Each of the stages is under the responsibility of their respective controller. The TestScenario controller (i.e. planner) accesses a prompt template library and generates the test cases. The library consists of a collection of prompts aimed at unveiling biases in LLMs, each of them specialized in a particular ethical concern. Every template has an associated test oracle, to evaluate whether an LLM output produces an acceptable response for such input prompt and Morales, [pg. 4, para. 3]; LangBiTe supports three different LLM providers: 0penAI, to prompt its proprietary LLMs, such as GPT-3.5 and GPT-4; HuggingFace Inference API, to access the Hugging Face hub hosted models; Replicate, a LLM hosting provider with further models not available on HuggingFace and further see Morales, [pg. 3, Fig. 1]; shows one or more tools (i.e. evaluator)).
receiving a response to each of the one or more requests; and (Morales, [pg. 3, Software Architecture]; (1) collects a subset of prompt templates from a prompt library as per the ethical concerns included; (2) for each prompt template, generates a test case addressing each of the sensitive communities selected; (3) executes the prompts against the LL Ms to evaluate; and ( 4) reports insights from the responses obtained from the LLM. The user must specify the number of templates to collect and the parameters to prompt LLMs as a test scenario).
While Morales teaches a test environment, Morales does not appear to explicitly teach to deployment. However Morales in view of the analogous art of Kotte (i.e. AI agent testing) does teach the entirety of the limitation: (Kotte, [02]; One or more embodiments provide benefits and/or solve one or more of the foregoing or other problems in the art with systems, methods, and non-transitory computer readable storage media by providing a contextual query answering system that trains and implements a unique machine learning architecture to generate accurate domain-specific contextual answers. In one or more embodiments, the disclosed systems train and utilize a context retrieval machine learning model to determine relevant contextual data for a contextual query and Kotte, [76]; In one or more embodiments, the context retrieval model utilizes transformers pretrained on the sentence similarity task (e.g., MP Net)).
While Morales teach evaluation of AI agent models and one or more tools, neither appear to explicitly teach: in response to the one or more requests, utilizing the planner to break at least one of the one or more requests into a plurality of sub-parts and to formulate a sequence of actions based on the plurality of sub-parts; and generating, by the agent core, an answer using the plurality of sub-parts with the memory module and the one or more tools; (Kotte, [65]; As part of the process for determining the relevant digital documents 416, the contextual query answering system 106 cleans and prepares query segments from the contextual query 402 by removing noise (such as special characters) and standardizing text. The contextual query answering system 106 also converts the preprocessed query segments into tokens (e.g., words, subwords, characters, and/or spaces). As shown, the contextual query answering system 106 passes the tokens into the encoder 404 and Kotte, [44]; the context retrieval model 208 is a machine learning model (e.g., a neural network) or a collection of machine learning models for generating or encoding query embeddings 210 from the contextual query 206 to represent the semantic content of the contextual query 206) and Kotte, [70]; As further shown in FIG. 4A, the contextual query answering system 106 performs a comparison to generate context scores 414 to determine relevant digital documents 416, such as relevant question-answer pairs 418. For example, the contextual query answering system 106 generates the context scores 414 representing a similarity (e.g., a cosine similarity or an embedding space distance) between the query embedding 406 and the data segment embeddings 412 and Kotte, [109]; the contextual query answering system 106 includes data storage manager 908. In particular, data storage manager 908 (implemented by one or more memory devices) stores the digital content used by the contextual query answering system 106 including the digital input, contextual query, embeddings, and contextual responses. The data storage manager 908 facilitates the generation of contextual responses by the contextual query answering system 106).
evaluating performance of the Al agent system based on the responses by determining (a) whether the responses sufficiently match expected responses for the one or more legitimate requests and (Kotte, [99]; Further, the contextual query answering system 106 performs the act 716 to compare the answer 706 with the predicted response 712. During the comparison, the contextual query answering system 106 uses performance metrics to assess how well the predicted response matches the ground truth answer. Based on the comparison, the contextual query answering system 106 performs the act 718 to adjust (or refine) the parameters for the response generator model 710 utilizing a loss function. For example, the contextual query answering system 106 analyzes discrepancies between the predicted response 712 and the answer 706 by performing a comparison to determine patterns in the differences).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the disclosed invention to have combined the teachings of Morales including evaluation of AI agent models and one or more tools with the teachings of Kotte including breaking prompts into subparts and evaluation of matching answers in order to provide the model with clean and standardized text and further to increase the accuracy of the model (Kotte, [65]; As part of the process for determining the relevant digital documents 416, the contextual query answering system 106 cleans and prepares query segments from the contextual query 402 by removing noise (such as special characters) and standardizing text. The contextual query answering system 106 also converts the preprocessed query segments into tokens (e.g., words, subwords, characters, and/or spaces) and Kotte, [99]; the contextual query answering system 106 analyzes discrepancies between the predicted response 712 and the answer 706 by performing a comparison to determine patterns in the differences. The contextual query answering system 106 iteratively repeats this process to finetune the response generator model 710 for increasing accuracy (or reduced loss) between predicted responses and ground truth answers over successive iterations).
While Morales teaches injecting an AI agent system with prompts and further testing the agent, Morales does not appear to explicitly teach malicious prompts. However, Morales in view of the analogous art of Kiciman (i.e. AI agent testing) does teach: providing the Al agent system with one or more requests including one or more legitimate requests and one or more malicious requests associated with techniques for compromising Al agents including prompt injection; (Kiciman, [pg. 7, Different positions of Attack Instructions]; For example, the number of test prompts for email QA is 11,250=50 (number of external content)×75(number of malicious instructions)×3(number of positions). The detailed statistical information of BIPIA is summarized in Table3. We have a total of 626,250 training prompts and 86,250 test prompts).
While Morales/Kotte teaches evaluating performance of an AI agent with respect to responses matching expected responses, Kotte does not appear to teach refraining from answering malicious prompts. However, Morales/Kotte in view of the analogous art of Kiciman (i.e. LLM performance) does teach the entirety of the limitation: …and (b) whether the Al agent system refrains from providing answers to the one or more malicious requests;… (Kiciman, [pg. 1, abstract]; In this work, we introduce the first benchmark, BIPIA, to measure the robustness of various LLMs and defenses against indirect prompt injection attacks. Our experiments reveal that LLMs with greater capabilities exhibit more vulnerable to indirect prompt injection attacks for text tasks, resulting in a higher attack success rate (ASR) for these attacks and Kiciman, [pg. 6, Different Positions of Attack Instructions]; The final prompt is obtained by injecting a malicious instruction collected at the attack level into an external content sample gathered at the task level using one of the three different positions and Kiciman, [pg. 7-8, Evaluation Metrics]; We use the attack success rate (ASR) as the primary metric to evaluate an LLM’s susceptibility to indirect prompt injection attacks. We design three approaches to verify the success of the attack for different attacks: rule based match, LLM-as-judge, and langdetect and Kiciman, [pg. 9, White-box Defense]; We propose a white-box defense method that applies adversarial training to the self-supervised fine-tuning stage of an LLM to teach it to ignore instructions in external content, thus enhancing its robustness against indirect prompt injection attacks).
storing results and performance metrics of the evaluation in a result store (Kiciman, [pg. 7, Table 4]; TABLE 4. ATTACK SUCCESS RATES (ASRS) OF DIFFERENT LLMS ON BIPIA. THE RESULTS ARE DISPLAYED IN DESCENDING ORDER OF LLM’S ELO RATING FROM CHATBOT ARENA and see table for visual representation of a store of evaluation metrics).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the disclosed invention to have combined the teachings of Morales/Kotte including injecting prompts and evaluating the performance of an AI agent system with the teachings of Kiciman including malicious prompts and further refraining from answering malicious prompts in order to ensure the AI agent does not deviate from user expectations due to lack of benchmark results (Kiciman, [pg. 1, Abstract]; These applications, however, are vulnerable to indirect prompt injection attacks, where malicious LLM misbehavior response instructions embedded within external content compromise LLM’s output, causing their responses to deviate from user expectations. Despite the discovery of this security issue, no comprehensive analysis of indirect prompt injection attacks on different LLMs is available due to the lack of a benchmark).
Further regarding Claim 8, while Morales/Kotte/Kiciman teaches a methods of Claim 1 (claim 8 rejected similarly), Morales does not appear to explicitly teach: A non-transitory computer-readable storage medium having computer-readable code stored thereon for programming one or more processors to perform steps of: However, Morales in view of the analogous art of Du (i.e. LLM functions) does teach the entirety of the limitation: (Kotte, [145]; The computing device 1200 includes a storage device 1206 includes storage for storing data or instructions. As an example, and not by way of limitation, the storage device 1206 can include a non-transitory storage medium described above. The storage device 1206 may include a hard disk drive (HDD), flash memory, a Universal Serial Bus (USB) drive or a combination these or other storage devices.).
One of ordinary skill in the art would have recognized that applying the known technique of Du would have yielded predictable results and resulted in an improved system. It would have been recognized that applying the technique of Du to the teachings of Morales would have yielded predictable results because the level of ordinary skill in the art demonstrated by the references applied shows the ability to incorporate determine performance of LLM and the use of computer readable media and cloud based systems. Further, applying computer readable media and cloud based systems to Morales would have been recognized by those of ordinary skill in the art as resulting in an improved system that allow for easier distribution of the system/product.
Further regarding Claim 15, while Morales/Kotte/Kiciman teaches a methods of Claim 1 (claim 15 rejected similarly), Morales does not appear to explicitly teach: Morales does not appear to explicitly teach: A cloud-based system comprising: one or more processors; and memory storing computer-executable instructions that, when executed, cause the one or more processors to: However, Morales in view of the analogous art of Kotte does teach the entirety of the limitation: (Kotte, [140-141]; A cloud-computing model can be composed of various characteristics such as, for example, on-demand self-service, broad network access, resource pooling, rapid elasticity, measured service, and so forth. A cloud-computing model can also expose various service models, such as, for example, Software as a Service (“SaaS”), Platform as a Service (“PaaS”), and Infrastructure as a Service (“IaaS”). A cloud-computing model can also be deployed using different deployment models such as private cloud, community cloud, public cloud, hybrid cloud, and so forth. In addition, as used herein, the term “cloud-computing environment” refers to an environment in which cloud computing is employed… Further, the computing device 1200 may be a server device that includes cloud-based processing and storage capabilities).
One of ordinary skill in the art would have recognized that applying the known technique of Du would have yielded predictable results and resulted in an improved system. It would have been recognized that applying the technique of Du to the teachings of Morales would have yielded predictable results because the level of ordinary skill in the art demonstrated by the references applied shows the ability to incorporate determine performance of LLM and the use of computer readable media and cloud based systems. Further, applying computer readable media and cloud based systems to Morales would have been recognized by those of ordinary skill in the art as resulting in an improved system that allow for easier distribution of the system/product.
Claims 5, 12, and 18 are rejected under 35 U.S.C. 103 as being unpatentable over Morales et al. "LangBiTe: A Platform for Testing Bias in Large Language Models", ARXIV.ORG, CORNELL UNIVERSITY LIBRARY, 201 OLINE LIBRARY CORNELL UNIVERSITY ITHACA, NY 14853, 29 April 2024 in view of Kotte et al. (US 20250252265 A1). And Kiciman, Emre et al.; “Benchmarking and Defending Against Indirect Prompt Injection Attacks on Large language Models,” Arvix.org, Association for Computing Machinery, 21 Dec 2023, and Du et al. (US 20250259005 A1) including provisional (63/657,694).
Regarding Claims 5, 12, and 18, While Morales/Kotte/Kiciman teaches performance of an AI agent and storing results, the cited references do not appear to teach additional requests based on results. However, Morales/Kotte/Kiciman in view of the analogous art Du (i.e. model testing) does teach the entirety limitation: The method of claim 1, wherein the steps further comprise: storing performance metrics of the Al agent system and providing the Al agent system with one or more additional requests based on its performance. (Du, [51], provisional [23]; The aforementioned iterative feedback loop also enables model self-healing, such that one or more LLMs will monitor the input provided and the resulting output to continuously learn and improve querying capabilities of the agent. In at least some embodiments, the feedback loop comprises a judge LLM providing and storing reasoning following assessment of the output of a prior LLM and Du, [73], provisional, [33]; an iterative loop is formed between query judge LLMs 109 and query LLM 106. In at least some embodiments, if any of the three or more query judge LLMs 109 (e.g. formatting LLM 110, correctness LLM 112, accuracy LLM 114) returns a query evaluation 118 of “fail”, query LLM 106 is instructed to update the query response 116 and provide an updated query response 116 to query judge LLMs 109 for further evaluation. The query LLM 106 can also receive the (failure) analysis (e.g. in the query prompt provided to the query LLM 106) and the score generated by the query judge LLMs 109 in order to improve the query response 116).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the disclosed invention to have combined the teachings of Morales including performance of an AI agent and see requests with the teachings of Du including storing and further requests in order to continue testing of models for further evaluation (Du, [73], provisional, [33]; an iterative loop is formed between query judge LLMs 109 and query LLM 106. In at least some embodiments, if any of the three or more query judge LLMs 109 (e.g. formatting LLM 110, correctness LLM 112, accuracy LLM 114) returns a query evaluation 118 of “fail”, query LLM 106 is instructed to update the query response 116 and provide an updated query response 116 to query judge LLMs 109 for further evaluation. The query LLM 106 can also receive the (failure) analysis (e.g. in the query prompt provided to the query LLM 106) and the score generated by the query judge LLMs 109 in order to improve the query response 116).
Examining Claims with Respect to Prior Art
Claims 2-4, 6-7, 9-11, 13-14, 16-17, and 20, though directed to non-statutory subject matter, are deemed to define over the currently known prior art under 35 USC 102 and 103. Examiner interprets based upon the claim limitations that there is no currently known prior art that discloses the features relating to: “The method of claim 1, wherein the steps further comprise: receiving a seed request comprising an expert curated question for a target application of the Al agent system and an associated expected response; generating one or more variations of the seed request via an Al-assisted amplification process configured to expand a question set based on the seed request while preserving an evaluation against the associated expected response; and providing the seed request and the one or more variations of the seed request to the Al agent system in the test environment for evaluating whether responses to the seed request and the one or more variations sufficiently match the associated expected response;” and “The method of claim 1, wherein the steps further comprise: scanning external feeds for examples of prompt injection; via a processing agent, scanning the web and the external feeds for the examples of prompt injection, wherein the external feeds include one or more third-party sources publishing LLM-related threat vectors; selecting an example of prompt injection as a seed request and generating one or more variations of the seed request; and providing the seed request and the one or more variations of the seed request to the Al agent system as at least a portion of the one or more malicious requests.”
The reason to withdraw the 35 USC 103 rejection of claims 2-4, 6-7, 9-11, 13-14, 16-17, and 20 in the instant application is because the prior art of record fails to teach the overall combination as claimed. Therefore, it would not have been obvious to one of ordinary skill in the art to modify the prior art to meet the combination above without unequivocal hindsight and one of ordinary skill would have no reason to do so. Upon further searching the examiner could not identify any prior art to teach these limitations. The prior art on record, alone or in combination, neither anticipates, reasonably teaches, not renders obvious the Applicant’s claimed invention.
Known Prior Art (patent)
US 20250252265 A1
GENERATING ANSWERS TO CONTEXTUAL QUERIES WITHIN A CLOSED DOMAIN
US 20250259005 A1
DATASET PARSING AND SEARCHING METHOD AND SYSTEM
US 20250190459 A1
SYSTEMS AND METHODS FOR DEVELOPMENT, ASSESSMENT, AND/OR MONITORING OF A GENERATIVE AI SYSTEM
US 20230359903 A1
Mitigation for Prompt Injection in A.I. Models Capable of Accepting Text Input
US 20250378386 A1
Managed Design and Generation of Artificial Intelligence Agents
US 20250103962 A1
SYSTEMS AND METHODS FOR GENERATING CUSTOMIZED AI MODELS
US 20240265114 A1
AN APPARATUS AND METHOD FOR ENHANCING CYBERSECURITY OF AN ENTITY
US 20250190461 A1
Method and System for Optimizing Use of Retrieval Augmented Generation Pipelines in Generative Artificial Intelligence Applications
US 20230205824 A1
Contextual Clarification and Disambiguation for Question Answering Processes
Known Prior Art (NPL)
Morales et al. "LangBiTe: A Platform for Testing Bias in Large Language Models", ARXIV.ORG, CORNELL UNIVERSITY LIBRARY, 201 OLINE LIBRARY CORNELL UNIVERSITY ITHACA, NY 14853, 29 April 2024
Known Prior Art (foreign)
CN116702146A
Injection vulnerability scanning method and system of Web server
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 JEREMY L GUNN whose telephone number is (571)270-1728. The examiner can normally be reached Monday - Friday 6:30-4:30.
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, Jerry O'Connor can be reached on (571) 272-6787. 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.
/JEREMY L GUNN/ Examiner, Art Unit 3624
Prosecution Insights