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 .
DETAILED ACTION
Note: Examiner Shaikh is now examining this application moving forward.
1. Claims 1, 4, 7, 11, 14 and 17 are amended.
2. Claims 1-20 are pending.
3 This office action is in response to the response filed on 12/23/25 for patent application number 18/809,140.
RESPONSE TO ARGUMENTS
Applicant argues#1
Turning now to the Action, Shirrell is relied on for teaching a predefined prompt modification according to prior-pending claims. However, Shirrell is limited to LLM retraining, and does not teach or fairly suggest the recited predefined prompt modification, much less the templates included in amended independent claim 1.
The Action argues that a "modified response" of Shirrell is equivalent to the recited predefined prompt modification. (Action at pg. 10, rejecting claim 4). Also or alternatively, the Action argues that LLM retraining is equivalent to the recited predefined prompt modification. (Action at pg. 8 (a ground truth response is used for retraining an LLM)). But neither the modified output of the LLM nor retraining of the LLM of Davison teaches or fairly suggests a predefined prompt modification. Quite simply, LLP outputs are not prompts, nor are LLM training operations.
Accordingly, Shirrell does not teach or fairly suggest "generat[ing] an output based on a prompt to an LLM, the prompt including open banking data and being generated from a first prompt generation template based on [a] difference between the output and [an] open banking objective, automatically identify[ing] a predefined training action and a predefined prompt modification based on the predefined prompt modification, identify[ing] or generat[ing] a second prompt generation template different from the first prompt generation template and generate a second prompt from the second prompt generation template, the second prompt including second open banking data; [and] generat[ing] a second output based on the second prompt to [a] retrained LLM," as recited in amended independent claim 1.
The remaining prior art of record does not remedy the deficiencies of Shirrell.
Accordingly, Applicant respectfully submits that the cited art does not reach Applicant's amended independent claim 1.
Amended independent claim 11 includes analogous limitations to those argued above, and therefore also benefits from those arguments. Dependent claims 2-10 and 12-20 depend, directly or indirectly, from one of independent claims 1 and 11, include additional limitations not taught or fairly suggested by the prior art, and are patentable for at least these reasons.
Accordingly, Applicant respectfully requests withdrawal of the pending prior art rejections of claims 1-20.
Examiner Response
Based on the amendments to the claims, the 35 U.S.C 103 rejection has been withdrawn.
Based on prior art search results, the prior art of record neither anticipate nor renders obvious the claimed subject matter of the instant application as a whole either taken alone or in combination, in particular, the prior art does not teach, “based on the predefined prompt modification, identifying or generating a second prompt generation template different from the first prompt generation template and generating a second prompt from the second prompt generation template, the second prompt including second open banking data; generating a second output based on the second prompt to the retrained LLM; and evaluating the second output with the objective function to determine a difference between the second output and the open banking objective”
The closest prior art of record:
1) Shirrell (US 2025/0284985) discloses, “Disclosed herein are system, method, and computer program product embodiments for using generative AI for prompt comparison. The system may receive a prompt. The prompt may include a request for information. The system may identify a stored prompt based on a similarity value between a received prompt for a large language model (LLM) and the stored prompt, the stored prompt including a response generated by the large language model (LLM). The system may generate a response using the LLM if the similarity value between the received and stored prompts is below a predefined threshold. The system may then modify the response by applying a first rule associated with a first designated phrase, and a second rule associated with a second designated phrase, where the first designated phrase comprises a banned phrase and where the second designated phrase comprises a selected phrase.”
2) US 2025/0278574 to Salim et al, discloses, “Aspects of this technical solution can receive, via a user interface, a first prompt for a large language model including a first query that references first data, generate one or more second prompts for the large language model based on the first prompt and the first data, each of the second prompts including one or more second data clarifying the first query, generate, by the large language model receiving one or more of the second prompts, one or more responses to the one or more second prompts, select an optimized prompt from among the one or more second prompts, according to a determination that a response to the at least one of the second prompts meets an accuracy threshold, and cause the user interface to present the optimized prompt or a response to the optimized prompt, the large language model to generate the response using the optimized prompt as input.”
3) US 2025/0371356 to Jia, discloses “Methods, systems, and computer-readable storage media for providing an initial version of a prompt template, the prompt template including dynamic input and first static input, generating a prompt using the initial version of the prompt template at least partially by populating the dynamic input with training data, receiving, from a large language model (LLM), an output that is responsive to the prompt, providing an evaluation at least partially based on the output, and selectively updating the prompt template to provide an updated version of the prompt template by prompting the LLM at least partially based on the evaluation, the updated version of the prompt template including second static input that is generated by the LLM and that is different from the first static input.”
Applicant argues#2
As described in the instant Specification:
The predefined training actions and predefined prompt modifications are preferably defined for a variety of system objectives and/or metrics relating thereto, a variety of differences between outputs and those objectives, and combinations thereof, and prescribe different types of training data, prompt modifications, and related parameters and combinations thereof.
[A] predefined training action may be more or less effective for moving the outputs and LLM toward the system objective(s) dependent on which prompt modification(s) are implemented therewith. Likewise, and for similar reasons, one or more of the predefined categories or characteristics (i.e., metadata) used for selection of a predefined prompt modification may depend at least in part on the training action(s) used in parallel therewith.
Advantageously, embodiments of the present invention implement iterative and parallel improvement of the LLM relative to given system objective(s) through automatically implemented prompt modification and fine-tuning efforts, including by learning relationships between predefined prompt modifications and fine-tuning or training data.
Paras [0069], [0076], [0078] (emphases added). Accordingly, amended independent claim 1 recites a second output based on a second prompt that is generated based on a second prompt generation template (that is different from a first generation template) and by a retrained LLM. In this manner, amended independent claim 1 recites "iterative and parallel improvement of the LLM relative to given system objective(s)" through parallel training and prompt modifications.
Recent Office guidance supports the eligibility of the present amended claims. For example, the Desjardins PTAB decision (Appeal 2024-000567), recently designated as precedential, emphasizes the importance of protecting artificial intelligence innovation where the technology at issue (logical structures and processes) improves training and/or operation of a machine learning model. This was true despite the fact that the Desjardins claims recited such an improvement at a relatively high level of abstraction (i.e., adjusting parameters in connection with optimizing for a first task while "protecting performance" in connection with performing a second task). At least in part relying on the Desjardins specification's description of corresponding benefits, comprising storage efficiencies and reduced system complexity, the claims were held to be eligible. Moreover, the PTAB warned that such cases should not be resolved under eligibility principles, but instead "that §§ 102, 103 and 112 are the traditional and appropriate tools to limit patent protection to its proper scope." (Pg. 10)
The present claims and specification present all the factors recommending eligibility under the reasoning of the Desjardins decision, perhaps even more compellingly. Under Desjardins, machine learning involving adjusting parameters to optimize for a first task while "protecting performance" for a second task was considered eligible despite being recited at a high level of abstraction. Here, the claims present more compellingly eligible (i.e., less abstract) machine learning-related operations, including retraining of an LLM based on an automatically curated training data set (according to a predefined training action) undertaken in parallel with an automatically determined and implemented predetermined prompt modification altering construction of a prompt for submission to the retrained LLM. Under Desjardins, the panel was compelled by description in the specification of benefits including "technical improvements over conventional systems [] addressing challenges in continual learning and model efficiency by reducing storage requirements and preserving task performance across sequential training." (Pg. 7) Here, the specification clearly describes "iterative and parallel improvement of the LLM relative to given system objective(s)" through the recited parallel training and prompt modifications. In various examples, the iterative and parallel operations permit "learning relationships between predefined prompt modifications and fine-tuning or training data," (para. [0078]) which is a technical benefit of the recited operations described in the specification.
Applicant respectfully submits that it is difficult to imagine a more analogous case to that which was decided under the Desjardins decision, and that the present case presents an even more compelling case for eligibility.
Amended independent claim 11 includes analogous limitations to those argued above, and therefore also benefits from those arguments. Dependent claims 2-10 and 12-20 depend, directly or indirectly, from one of independent claims 1 and 11, include additional limitations contributing to the eligibility thereof, and are eligible for at least these reasons.
Examiner Response
Examiner respectfully disagrees.
Examiner reproduces paras 69,76, 78 and other paras of the specification below:
[0064] In one or more embodiments, the LLM is initially a licensed model such as those made available under the trademarks GPT-4 or CHATGPT® (registered trademarks ofOpenAI OpCo, LLC) as of the date of the initial filing of the present disclosure. As discussed above, the LLM may be hosted in a cloud computing environment, locally on one or more service devices, or otherwise within the scope of the present invention. The generalized training or pretraining of the LLM in the state received under a license, for example- may be conducted on a wide variety of data, according to known practices associated with commercially available models such as those listed above
[0065] A preliminary step according to embodiments of the present invention may include selecting an LLM trained on language expected to be predominantly encountered in prompts and fine-tuning data corresponding to system objectives and/or the open banking service and/or region in which the open banking service will predominantly be provided. For example, the LLM may be trained on English, French, Spanish, German, Mandarin, Cantonese, Arabic, Hindi or other languages, and may be more particularly trained on data filtered for region or ethnicity (e.g., American English or British English), financial channel and/or other differences. Because the LLM will mostly encounter and output language unique to conversations reliant on particularized open banking data and vernacular, it is also foreseen that additional filters such as for language used in economic, financial and transactional contexts- may additionally be applied to select LLMs trained on particularly relevant language within the scope of the present invention.
[0066] Further, fine-tuning of the LLM may be initially performed (e.g., before being placed in a production environment for open banking services) in view of the open banking service and system objective(s) for which it is to be used. In one or more embodiments, fine-tuning includes one or more of self-supervised, supervised and/or reinforcement learning. As discussed in more detail above, the fine-tuning may be performed with data types including open banking data such as memo or description fields of open banking records, FI data regarding data subjects and/or FI terms and rate limits, transaction and credit card data, account data, firmographic entity data, location data, value data, regulatory data, PII, entity identification and/or authentication data, and/or other financial and relevant data, and combinations thereof.
[0067] Similarly, the initially fine-tuned LLM may be tested with one or more prompts, preferably engineered according to one or more predefined strategies, to determine capabilities, efficiency and accuracy with reference to the open banking service and system objective(s) for which it is to be used. Where open banking data is submitted with the test prompts, it likewise may originate with or include memo or description fields of open banking records, FI data regarding data subjects and/or FI terms and rate limits, transaction and credit card data, account data, firmographic entity data, location data, value data, regulatory data, PII, entity identification and/or authentication data, and/or other financial and relevant data, and combinations thereof
[0069] The predefined training actions and predefined prompt modifications are preferably defined for a variety of system objectives and/or metrics relating thereto, a variety of differences between outputs and those objectives, and combinations thercof, and prescribe different types of training data, prompt modifications, and related parameters and combinations thereof
[0076] In other words, a predefined training action may be more or less effective for moving the outputs and LLM toward the system objective(s) dependent on which prompt modification(s) are implemented therewith. Likewise, and for similar reasons, one or more of the predefined categories or characteristics (i.e., metadata) used for selection of a predefined prompt modification may depend at least in part on the training action(s) used in parallel therewith
[0078] Advantageously, embodiments of the present invention implement iterative and parallel improvement of the LLM relative to given system objective(s) through automatically implemented prompt modification and fine-tuning efforts, including by learning relationships between predefined prompt modifications and fine-tuning or training data
Examiner submits that fine tuning a large learning model (LLM) is a common technique that is used in industry to customize a specific model for a specific task.
Furthermore, modifying a prompt is commonly used technique by developers to interact and control the LLMs.
The above cited paras of the spec disclose the LLM recited at a high level of generality and operating in its ordinary capacity, as is being used as a tool to implement the steps of the identified abstract idea, see MPEP 2106.05(f).
Applicants’ claims do not improve technology; the underlying technology remains unaffected by the claims. Applicant is merely using existing technology (for its intended purpose) to implement the identified abstract idea. Any improvements lie in the abstract idea itself, not in underlying technology.
The claims of the Ex Parte Desjardins decision analyzed eligibility to determine whether
the claims were directed to an improvement in the functioning of the computer or an
improvement to other technology or technical field. It was in step 2a prong 2, it was
determined that the specification identified improvements and was reflected in the
claims as to how the machine learning model itself operates.
The specification of the Desjardins application identified the improvement to machine
learning technology.
Whereas the claims and the specification of the instant application, do not reflect the
improvement to the machine learning models (the claimed LLM).
Therefore, the claims are unlike the claims in Ex Parte Desjardins.
The rejection is maintained.
Claim Rejections - 35 USC § 101
1. 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.
2. Claims 1-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more. A Section 101 analysis is below.
3. Step 1- are the claims directed to a process, machine, manufacture or composition of matter.
The CRM of claim 1 and method of claim 11 are within the statutory categories of invention.
Step 2A, prong one - do the claims recite a judicial exception, which is an abstract idea
enumerated in MPEP 2106, a law of nature, or a natural phenomenon. Using the text of claim 11 as an example, claims 1 and 10 recite:
11. A computer-implemented method for providing dynamic large language model
(LLM) open banking services, comprising, via one or more transceivers and/or processors:
generating an output based on a prompt to an LLM, the prompt including open banking data and being generated from a first prompt generation template;
evaluating the output with an objective function to determine a difference between the output and an open banking objective;
based on the difference between the output and the open banking objective, automatically identifying a predefined training action and a predefined prompt modification;
based on the predefined training action, curating a training data set and retraining the LLM on the training data set to generate a retrained LLM;
based on the predefined prompt modification, identifying or generating a second prompt generation template different from the first prompt generation template and
generating a second prompt, from the second prompt generation template, the second prompt including second open banking data;
generating a second output based on the second prompt to the retrained LLM; and
evaluating the second output with the objective function to determine a difference between the second output and the open banking objective.
Referring to the bolded limitations above, independent claims 1 and 11 recite an abstract idea enumerated in MPEP 2106. Specifically, as drafted claims 1 and 11 only recite the abstract idea of a mental process of generating an output based on a prompt, the prompt including open banking data. Please see MPEP 2106.04(a)(2)(III) which notes claims do recite a mental process when they contain limitations that can
practically be performed in the human mind, including for example, observations, evaluations, judgments, and opinions. Examples of claims that recite mental processes include: a claim to "collecting information, analyzing it, and displaying certain results of the collection and analysis". Accordingly, claims 1 and 11 are directed to the judicial exception of an abstract idea.
Step 2A, prong two - do the claims recite additional elements that integrate the judicial
exception into a practical application. Integration of the judicial exception into a practical application requires an additional element or a combination of additional elements in the claim to apply, rely on, or use the judicial exception in a manner that imposes a meaningful limit on the judicial exception, such that the claim is more than a drafting effort designed to monopolize the exception.
Regarding claims 1 and 11, these claims only recite the additional elements of a computer, large language model (LLM), one or more transceivers and/or processors, and retrained LLM to perform a method for providing dynamic open banking services, comprising: an objective function; based on the predefined training action, curating a training data set and retraining on the training data set to generate a retrained LLM.
The components are recited at a high-level of generality (e.g., to receive, store, or transmit data) such that it amounts no more than mere instructions to apply the exception using generic components. Accordingly, the additional elements do not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea. Please see MPEP 2106.05(f)(1) discussing when the claim recites only the idea of a solution or outcome i.e., the claim fails to recite details of how a solution to a problem is accomplished this does not show integration into a practical application. Please see MPEP 2106.05(f)(2) discussing when the claim invokes computers or other machinery merely as a tool to perform an existing process including use of a computer or other machinery for economic tasks this does not show integration into a practical application. It is further noted that the claimed invention as recited in claims 1 and 11 do not pertain to an improvement in the functioning of the computer components themselves or a technological solution to a technological problem. Regarding the LLM additional element, please also see Recentive Analytics, Inc. V. Fox Corp., No. 2023-2437 (Fed. Cir. Apr. 18, 2025), which affirmed the District of Delaware's dismissal of Recentive's infringement suit on the ground that the asserted patents were directed to ineligible subject matter under 35 U.S.C. § 101. The decision reinforces the courts' view that applying generic machine learning techniques to known problems - without technical innovation in the machine learning methods themselves - is insufficient
for patent eligibility. It is respectfully noted that the claimed LLM is a third party product, as noted in Applicant's specification, [0064].
Step 2B ..... do the claims recited additional elements that amount to significantly more than the judicial exception. Regarding claims 1 and 11, as discussed above with respect to integration of the abstract idea into a practical application, the additional elements of a computer, large language model (LLM), one or more transceivers and/or processors, and retrained LLM to perform a method for providing dynamic open banking services, comprising: an objective function; based on the predefined training action, curating a training data set and retraining on the training data set to generate a retrained LLM are no more than mere instructions to apply the exception using generic computer components. Mere instructions to apply an exception using a generic computer component cannot provide an inventive concept. As noted above, the LLM is a third party product hosted in a computing environment. Please see Applicant's
specification, [0064], [0020]. Accordingly, claims 1 and 11 do not recite additional elements that amount to significantly more than the judicial exception.
In view of the above analysis, independent claims 1 and 11 are not patent eligible. Dependent claims 2-10 and 12-20 do not cure the deficiencies in their respective base claims. Specifically, claims 2-10 and 12-20 merely refine the abstract idea (2A1) by invoking a computer as a tool to perform an existing process (2A2, 2B). Regarding the further additional element in the dependent claims including prompt engineering modification, a prompt architecture modification, or a multi-shot learning
modification (claims 4, 14); second retrained LLM (claims 7, 17); complexity platform comprising one or more non-linear, recursive, and super literal genetic algorithms and one or more prompt engineering heuristics (claims 9, 19), please see MPEP 2106.05(f)(2) discussing when the claim invokes computers or other machinery merely as a tool to perform an existing process including use of a computer or other
machinery for economic tasks this does not show integration into a practical application or provide significantly more.
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 MOHAMMAD Z SHAIKH whose telephone number is (571)270-3444. The examiner can normally be reached M-T, 9-600; Fri, 8-11, 3-5.
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, BENNETT SIGMOND can be reached at 303-297-4411. 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.
/MOHAMMAD Z SHAIKH/Primary Examiner, Art Unit 3694 8/18/2026
Prosecution Insights