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 .
Drawings
The drawing submitted on 06/05/2024 is considered by the examiner.
Response to Amendment
Claims 1, and 3-21, are currently pending in the application and among them claims 1, and 15-16, are independent claims and claims 1, 3-5, 7, 15-16 has been amended and claims 17-21 has been added as new.
Response to Arguments
Applicant's arguments filed on 05/19/2026 have been fully considered but they are not persuasive. The following are the examiner response with respect to applicant arguments:
35 U.S.C. & 101 Rejections
Applicant Argument 1: The Examiner is respectfully reminded of the "Advance notice of change to the MPEP in light of Ex Parte Desjardins" issued on December 5, 2025 (hereafter, "December 2025 Notice"), and the Appeals Review Panel decision in Ex Parte Desjardins, Appeal No. 2024-000567 (PTAB September 26, 2025, Appeals Review Panel Decision) designated precedential on November 4, 2025. As discussed in the December 2025 Notice, eligibility of claims may be dependent on whether they are directed to and improvement in the functioning of a computer, or an improvement to other technology or technical field.
Even if the claims were directed towards a judicial exception, which the Applicant does not concede, the alleged judicial exception is integrated into a practical application in accordance with MPEP 2106.04. The Revised 2025 Guidelines set out a two-part examination. "[F]irst, the specification should be evaluated to determine if the disclosure provides sufficient details such that one of ordinary skill in the art would recognize the claimed invention as providing an improvement in the functioning of a computer, or an improvement to other technology or a technical field."
The Revised 2025 Guidelines state that in a second step, "if the specification sets forth an improvement in technology or in a technical field, the claim must be evaluated to ensure that the claim itself reflects the disclosed improvement, i.e., the claim includes the components or steps of the invention that provide the improvement described in the specification." Here, the Applicant's claims are directly reflective of the disclosed improvement. For example, among other elements the claims recite "the data characterizing the second prompt modifies the received data characterizing the first prompt based on at least one of a type of the artificial intelligence based model, a setting of the artificial intelligence based model, or a configuration for an enterprise in which the artificial intelligence based model is deployed.” In other words, the claims directly reflect the technical benefits described above.
For at least these reasons, withdrawal of the pending rejections under 35 U.S.C. § 101 is respectfully requested.
Examiner Response 1: Examiner agrees with the applicant with reference to Ex Parte Desjardins ground of patentability if claimed invention as providing an improvement in the functioning of a computer, or an improvement to other technology or a technical field.", and "if the specification sets forth an improvement in technology or in a technical field, the claim must be evaluated to ensure that the claim itself reflects the disclosed improvement, i.e., the claim includes the components or steps of the invention that provide the improvement described in the specification."
However pending claims only reflects the improvement of the mental process which is improvement of a prompt by modifying the firs prompt to a second prompt corresponding to an artificial intelligence base model for generating correct response from the artificial base model. The claims do not show any improvement on neither the artificial intelligence base model nor the technology. Further technology is not improved evidence by rejection of the claims. This improvement to the prompt is tied to improvement to the abstract idea not specifically the technology behind it. An AI model will always present a response dependent on the prompt as this is what AI always do. So, the argument of improving the prompt improves the answer would not help overcome the 101 since a human can mentally optimize the prompt and in return the model will provide a different answer based on its input. Therefore, applicant’s argument with respect to 101 rejection based on Exparte Desjardin is not persuasive and the rejection of all claims remain same.
35 U.S.C. & 102 Rejections
Applicant Arguments 2: Almaer does not teach outputting a response in natural language as recited in the claims. Additionally, Almaer is entirely silent as to generating a second prompt or enhanced prompt based on characteristics of an artificial intelligence model.
Examiner Response 2: Examiner respectfully disagree with applicant’s piecemeal analysis and citation of prior art teaching to support applicant’s argument without disclosing other teaching paragraphs that reflects the teaching of the claimed limitation.
First of all, Examiner would like to point out to applicant that, nowhere in the claim recites applicant argument with respect to “enhanced prompt based on characteristics of an artificial intelligence model” instead it recites “wherein the data characterizing the second prompt modifies the received data characterizing the first prompt based on at least one of a type of the artificial intelligence based model” which has completely different meaning.
With respect to the prior art teaching the office action clearly state that if the prompt structure is not properly structured for the LLM model input, the prompt is modified to a prompt structure till the prompt generates a response from LLM model ([0030] The data requests (e.g., queries) directed at an endpoint are required to be compliant with various requirements associated with the endpoint. In particular, query statements for an endpoint must be structured to comply with both the syntax of a query language and any requirements which may be stipulated for queries that are supported by an API (or other software interface) for the endpoint.[0035] The system may match a user request (e.g., a data retrieval request) to a “best” prompt template, out of a set of such templates, for an LLM. A prompt template may, for example, comprise a previous query that was accepted by the endpoint or an example of a properly constructed query for the endpoint. The matched template may then be provided in an input prompt to the LLM with instructions to generate a query for submitting to the endpoint. [0036] When a user provides a first data request (expressed using natural language) for an endpoint, the system is configured to instruct an LLM to generate a query for the endpoint, i.e., by converting the first data request to a corresponding query. [0039] The retrieved previous query is provided, along with the first data request, as input to the LLM. In particular, the first data request and the retrieved previous query may be included as part of an input prompt to the LLM, with instructions for the LLM to generate a query for the endpoint. In this way, the LLM may be provided with a “best” example of a desired output, i.e., a query that captures the user's request and that is compliant with the requirements of the endpoint, when generating the query.).
The examiner interpreted Syntex and query language as data characterizing prompt. Further Almer teaches prompt modification due to error response and modify the prompt structure to receive correct results. Examiner also interpreted that as prompt with greater relevancy to LLM model will only generate the corrected response ([0040] The system may then provide the generated query to the endpoint. If the system receives an error response from the endpoint, details about the error, i.e., error data, may be obtained and provided as further input to the LLM. The error data may specify, for example, information on the line number, character number, etc. that contains the erroneous code. The error data is provided as part of an input prompt to the LLM to re-generate the query for the endpoint. Other feedback from the endpoint may be additionally provided to the LLM in the instruction to re-generate the query. [0041] This process of instructing the LLM to generate a query corresponding to the first data request based on modifying an input prompt to the LLM may proceed iteratively until a successful response is received from the endpoint. Upon determining a successful query, i.e., a query that is accepted by the endpoint, the system may update a knowledge base, such as a queries database, storing query information of queries for the endpoint.).
Almaer et al. further teach the claimed limitation of “data characterizing the second prompt modifies the received data characterizing the first prompt based on at least one of a type of the artificial intelligence based model, a setting of the artificial intelligence based model, or a configuration for an enterprise in which the artificial intelligence based model is deployed’ in paragraph [0028] The output/result may be generated by the generative AI model through interpreting the intent and context of the prompt. In some cases, the generative AI model may be implemented with constraints on the acceptable prompts. In some cases, this may include a prompt template. A prompt template may specify that prompts have a certain structure or constrained intents, or that acceptable prompts exclude certain classes of subject matter or intent, such as the production of results or outputs that are violent, pornographic, etc. [0030] An endpoint may implement (or expose) a software interface, such as an application programming interface (API), for offering various services to other computer programs. An API contains and is implemented by function calls, which are language statements that request software to perform particular actions and services. The specification of the API describes functions and other parameters that are supported by the API. The data requests (e.g., queries) directed at an endpoint are required to be compliant with various requirements associated with the endpoint. In particular, query statements for an endpoint must be structured to comply with both the syntax of a query language and any requirements which may be stipulated for queries that are supported by an API (or other software interface) for the endpoint. [0044] To illustrate additional details regarding the methods and systems of the present application, some concepts relevant to generative AI models, neural networks, and machine learning (ML) are first discussed. Therefor Almaer et al. clearly teach the claimed limitation of “ generating data characterizing a second prompt, wherein the second prompt is configured to generate a response from an artificial intelligence model that has a greater relevancy than a response from the artificial intelligence model generated by providing the first prompt to the artificial intelligence model, wherein the data characterizing the second prompt modifies the received data characterizing the first prompt based on at least one of a type of the artificial intelligence based model, a setting of the artificial intelligence based model, or a configuration for an enterprise in which the artificial intelligence based model is deployed.
Applicant arguments are therefore not persuasive and rejection of all claims remain same.
35 U.S.C. & 103 Rejections
Applicant Argument 3: Tsun does not cure the deficiencies of Almaer. Tsun's prompt generation is based on user attributes, user connections, and messaging intent derived from historical activity, not on the technical characteristics or deployment configuration of the AI model. Since neither Almaer nor Tsun, alone or in combination, teaches all of the elements of the independent claims, the combination cannot render the amended claims obvious.
Examiner Response 3: Examiner respectfully disagree with applicant’s simple assertion in conclusion of the Tsun’s not teaching the claims 11-13.
First of all, claims nowhere recites the underline applicant argument. “data characterizing modifying first prompt to second prompt based on the technical characteristics or deployment configuration of the AI model rather it recites data characterizing modifying first prompt to second prompt based on at least type of AI based model, a settings of the AI based model or a configuration for an enterprise in which the AI based is deployed. Almaer et al. teaches that by teaching the prompt modification based on prompt supported by the API associated LLM ([0030], [0035-0036] and [0039])which examiner interpreted as modifying first prompt to second prompt based on at least type of AI based model, a settings of the AI based model or a configuration for an enterprise in which the AI based is deployed.
Tsun’s teaching is in the same field of technology, and Tsun’s teaching was combined to teach the deficiency of Almaer et al. for claim 11, which is modifying the first prompt into a second prompt based on generating a blueprint (a second subset of prompt inputs per Tsun teaching) corresponding to a user behavior including historical data analysis characteristics. Tsun clearly teaches the limitation of claim 11, in paragraphs, [0044] In some embodiments, prompt generation component 160 determines the messaging intent based on historical activity data of the user of user system 110. For example, prompt generation component 160 determines that the messaging intent is to seek work if the user of user system 110 has recently applied to one or more jobs. [0046] In some embodiments, prompt generation component 160 maps a set of user attributes to a set of one or more prompt inputs using the connection. [0048] In some embodiments, input generation component 164 creates an initial prompt using a first subset of prompt inputs of the set of prompt inputs mapped to the user attributes and updating the initial prompt to generate prompt 106 which includes a second subset of prompt inputs of the set of prompt inputs. This may be useful when the generated suggestion can have two different styles. Input generation component 164 updates the initial prompt to include additional information from attribute data 104. For example, input generation component 164 includes honors data 332 and certifications data 328. This updated portion of the prompt including honors data 332 and certification data 328 can result in suggestions 114 that read in a list format. By generating the prompts for these separately, content generation system 100 ensures that the resulting suggestions 114 include both writing styles where necessary. The terms set of prompt inputs, first subset of prompt inputs, and second subset of prompt inputs as used in this disclosure can be a single prompt input or multiple prompt inputs. [0052] Prompt generation component 160 creates prompt 106, x, based on the extracted attribute data 104. In some embodiments, prompt generation component 160 creates more than one prompt. [0053] For each prompt 106, x, the deep learning model 108 produces one or more outputs y and, for each output y, a score P (x, y) that indicates a likelihood of the prompt x and the respective output y occurring together. Using the output(s) y and corresponding score(s) P (x, y), the deep learning model 108 generates first versions of one or more suggestions 114. The first versions of the one or more suggestions 114 each include at least one piece of writing that has been machine-generated by the deep learning model 108.
With respect to claim 12, Tsun teaches the limitation of “wherein the blueprint is at least partially automatically generated based on metadata ((a set of user attributes to a set of one or more prompt inputs), [0046] In some embodiments, prompt generation component 160 maps a set of user attributes to a set of one or more prompt inputs using the connection.).
With respect to claim 13, Tsun teaches the limitation of “wherein generating data characterizing the second prompt is based at least on user feedback to historical provided prompt responses”( [0044] In some embodiments, prompt generation component 160 determines the messaging intent based on historical activity data of the user of user system 110. For example, prompt generation component 160 determines that the messaging intent is to seek work if the user of user system 110 has recently applied to one or more jobs. [0045] In some embodiments, prompt generation component 160 maps a set of user attributes to a set of one or more prompt inputs using the identifier. For example, prompt generation component 160 maps user attributes that are relevant and effective to display for a user with entry level experience (e.g., education) while excluding user attributes that are irrelevant and ineffective to display for a user with entry level experience (e.g., years of experience). [0048] In some embodiments, input generation component 164 creates an initial prompt using a first subset of prompt inputs of the set of prompt inputs mapped to the user attributes and updating the initial prompt to generate prompt 106 which includes a second subset of prompt inputs of the set of prompt inputs. [0059] In some embodiments, deep learning model 108 sends suggestion 114 to prompt feedback component 168 of prompt generation component 160. Prompt feedback component 168 is a component that receives suggestion 114 from deep learning model 108 and feedback 116 from user system 110 and uses them to generate future prompts. For example, prompt feedback component 168 generates updated prompts based on suggestions 114 and/or feedback 116. [0061] In some embodiments, prompt feedback component 168 receives feedback from user system 110. For example, user system 110 includes user interface 112 and, as explained with reference to FIGS. 4-9, user interface 112 includes a graphical user interface (such as graphical user interface 400 of FIGS. 4-9). The graphical user interface can include a profile interface displaying suggestion 114 with which a user can interact. For example, the profile interface displays suggestion 114 and the user interacts with the profile interface to refresh the suggestion. In response to receiving this interaction, user system 110 sends feedback 116 to prompt feedback component 168, indicating that the suggestion should be refreshed.
According to MPEP 2141.01(a) I., “A reference is analogous art to the claimed invention if: (1) the reference is from the same field of endeavor as the claimed invention (even if it addresses a different problem); or (2) the reference is reasonably pertinent to the problem faced by the inventor (even if it is not in the same field of endeavor as the claimed invention). Note that "same field of endeavor" and "reasonably pertinent" are two separate tests for establishing analogous art; it is not necessary for a reference to fulfill both tests in order to qualify as analogous art. See Bigio, 381 F.3d at 1325, 72 USPQ2d at 1212. The examiner must determine whether a reference is analogous art to the claimed invention when analyzing the obviousness of the subject matter under examination. When more than one prior art reference is used as the basis of an obviousness rejection, it is not required that the references be analogous art to each other. See Sanofi-Aventis Deutschland GMbH v. Mylan Pharms. Inc., 66 F.4th 1373, 1380, 2023 USPQ2d 552 (Fed. Cir. 2023).
Since Tsun et al. is analogous art also teach the deficiency part of the limitation of Almaer et al. therefore basis of the obviousness is proper as per MPEP. Therefore, applicant arguments with respect to Tsun et al. teaching is not persuasive and all claims rejection remain same.
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.
Claims1 and 15-16, are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
The claim(s) recite(s), receiving data characterizing a first prompt from a user interface; generating data characterizing a second prompt, wherein the second prompt is
configured to generate a response from an artificial intelligence model that has a greater relevancy than a response from the artificial intelligence model generated by providing the first prompt to the artificial intelligence model, wherein the data characterizing the second prompt modifies the received data characterizing the first prompt based on at least one of a type of the artificial intelligence based model, a setting of the artificial intelligence based model, or a configuration for an enterprise in which the artificial intelligence based model is deployed; receiving data characterizing a response to the second prompt by providing the data characterizing the second prompt to an artificial intelligence based model; and providing the response to the second prompt in the user interface. The limitation, as drafted is a process that, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for recitation of generic computer components. That is, other than reciting “artificial intelligence model” and “artificial intelligence base model” nothing in the claim element precludes the step from practically being formed in the mind. For example, but for the “artificial intelligence model” and “artificial intelligence base model” language “receive”, “generating”, “modify” and “providing” in the context of this claims encompasses a person verbally receiving from another person an initial request to find location information related to an address in a way that is not clear to the person from the other person initial request and upon rephrasing the user initial request into another request that the person understand clearly, the person provide the other person information for the request.
If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it fall with the “Mental Process” grouping of abstract ideas. Accordingly, the claim recites an abstract idea.
The judicial exception is not integrated into a practical application. In particular, the claim recites addition elements -receiving data characterizing a response to the second prompt by proving the data characterizing the second prompt to an artificial intelligence based model. The use of “artificial intelligence base model” to receive the response to the second prompt, is recited at a high-level of generality (i.e. receiving by a computing device a response/result to a text input/natural language request) such that it amounts no more than mere instructions to apply the exception using generic computer components.
Accordingly, this additional elements does not integrate the abstract into a practical application because it does not impose any meaningful limits on practicing the abstract idea. The claim is directed to an abstract idea.
The claims do 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 element of receiving data characterizing a response to the second prompt by proving the data characterizing the second prompt to an artificial intelligence based model, steps amounts to no more than mere instructions to apply the exception using a generic computer component. With respect to use of a generic computer component to receive content or information, the courts have indicated that “finding the use of a generic server insufficient to add inventive concepts to an abstract idea” (See MPEP 2106.05 (a) I., Particular structure of a server that stores organized digital images, TLI Communications, 823 F.3d at 612, 118 USPQ2d at 1747). Also, court has indicated that “delivering broadcast content to a portable electronic device such as a cellular telephone, when claimed at a high level of generality” may not be sufficient to show an improvement to technology (See MPEP 2106.05 (a) Il., Affinity Labs of Tex. v. Amazon.com, 838 F.3d 1266, 1270, 120 USPQ2d 1210, 1213 (Fed. Cir. 2016); Affinity Labs of Tex. v. DirecTV, LLC, 838 F.3d 1253, 1262, 120 USPQ2d 1201, 1207 (Fed. Cir. 2016)).
Further the limitations receiving data characterizing a response to the second prompt by proving the data characterizing the second prompt to an artificial intelligence based model, provide nothing more than mere instructions to implement an abstract idea on a generic computer component. The use of artificial intelligence base model, as the claim recites, provide only the idea of a solution or outcome and fails to recite details of how a solution to a problem is accomplished. Without any description of the Al mechanism for accomplishing the result using generic processor, does not integrate a judicial exception into a practical application or provide significantly more because this type of recitation is equivalent to the words "apply it" (See MPEP 2106.05(f), “The recitation of claim limitations that attempt to cover any solution to an identified problem with no restriction on how the result is accomplished and no description of the mechanism for accomplishing the result, does not integrate a judicial exception into a practical application or provide significantly more because this type of recitation is equivalent to the words "apply it". See Electric Power Group, LLC v. Alstom, S.A., 830 F.3d 1350, 1356, 119 USPQ2d 1739, 1743-44 (Fed. Cir. 2016); Intellectual Ventures | v. Symantec, 838 F.3d 1307, 1327, 120 USPQ2d 1353, 1366 (Fed. Cir. 2016): Internet Patents Corp. v. Active Network, Inc., 790 F.3d 1343, 1348, 115 USPQ2d 1414, 1417 (Fed. Cir. 2015). In contrast, claiming a particular solution to a problem or a particular way to achieve a desired outcome may integrate the judicial exception into a practical application or provide significantly more. See Electric Power, 830 F.3d at 1356, 119 USPQ2d at 1743.”). Mere instructions to apply an exception using a generic computer component cannot provide an inventive concept. The claim is not patent eligible.
Even when considered in combination, the additional elements represents mere instructions to apply an exception and insignificant extra-solution activity which cannot provide an inventive concept. Claims 1 and 15-16 are thus patent ineligible.
With respect to Claims 6-14, similar to claim 1, nothing in the claim element precludes the step from practically being formed in the mind. If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it fall with the “Mental Process” grouping of abstract ideas. Accordingly, the claims recites an abstract idea.
The claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claims are not patent eligible.
Even when considered in combination, the additional elements represents mere instructions to apply an exception and insignificant extra-solution activity which cannot provide an inventive concept. Claims 6-14 are thus patent ineligible.
Claim Rejections - 35 USC § 112
The following is a quotation of the first paragraph of 35 U.S.C. 112(a):
(a) IN GENERAL.—The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor or joint inventor of carrying out the invention.
The following is a quotation of the first paragraph of pre-AIA 35 U.S.C. 112:
The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor of carrying out his invention.
Claims 17-18 are rejected under 35 U.S.C. 112(a) or 35 U.S.C. 112 (pre-AIA ), first paragraph, as failing to comply with the written description requirement. The claim(s) contains subject matter which was not described in the specification in such a way as to reasonably convey to one skilled in the relevant art that the inventor or a joint inventor, or for applications subject to pre-AIA 35 U.S.C. 112, the inventor(s), at the time the application was filed, had possession of the claimed invention. The claim recites limitation “wherein generating the data characterizing the second prompt further comprises: modifying the received data characterizing the first prompt to restrict a response of the artificial intelligence based model to information contained within a provided dataset object” which is not supported by the specification [0045] including the incorporated by reference patent application 16/512647 and provisional application 63/471249.
Claim 18 depends on claim 17 and include all the limitation of claim 17 and thus rejected based on the dependency on the rejected base claim 17.
For the purpose of the examination examiner will reject the claim 17, based on a prior art teaching.
Claim Rejections - 35 USC § 102
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(2) the claimed invention was described in a patent issued under section 151, or in an application for patent published or deemed published under section 122(b), in which the patent or application, as the case may be, names another inventor and was effectively filed before the effective filing date of the claimed invention.
Claim(s) 1-10,14-17 and 19-20, are rejected under 35 U.S.C. 102(a)(2) as being anticipated by Almaer et al.(US 2024/0362209 A1).
Regarding Claims 1 and 15-16, Almaer et al. teach: A method comprising: ([0023] In another aspect, the present application discloses a computing system. The computing system includes a processor and a memory coupled to the processor. The memory stores computer-executable instructions that, when executed by the processor, may cause the processor to: receive a request for retrieval of data satisfying one or more criteria, the request including at least one data request parameter; search a database storing example queries based on the request to identify at least one matching query; provide, to a large language model (LLM), an input prompt to generate a query purporting to retrieve data satisfying the one or more criteria, the input prompt including the at least one data request parameter and the at least one matching query as an example; and receive, from the LLM, a result including the generated query.): receiving data characterizing (constructed query) a first prompt (input prompt ) from a user interface ([0034] It is desired to provide a process for automatically generating queries for an endpoint that are based on user requests and that comply with the requirements of the endpoint. [0035] The system may match a user request (e.g., a data retrieval request) to a “best” prompt template, out of a set of such templates, for an LLM. A prompt template may, for example, comprise a previous query that was accepted by the endpoint or an example of a properly constructed query for the endpoint. The matched template may then be provided in an input prompt to the LLM with instructions to generate a query for submitting to the endpoint. [0077] In some implementations, the data request may be received via a user interface on the user device 120.); generating data characterizing (previous query is provided, along with the first data request ) a second prompt (modifying an input prompt to the LLM), wherein the second prompt is configured to generate a response from an artificial intelligence model (LLM) that has a greater relevancy (similarity or embeddings associated with all or a subset (e.g., only correctly formed queries) of previous data) than a response from the artificial intelligence model generated by providing the first prompt to the artificial intelligence model; receiving data characterizing a response (a result including the generated query) to the second prompt by providing the data characterizing the second prompt to an artificial intelligence based model (an API endpoint associated with a third-party server) ([0028] In the present application, the term “generative AI model” may be used to describe a machine learning model. A generative AI model may sometimes be referred to, or may use, a language learning model. [0030] An endpoint may implement (or expose) a software interface, such as an application programming interface (API), for offering various services to other computer programs. An API contains and is implemented by function calls, which are language statements that request software to perform particular actions and services. The specification of the API describes functions and other parameters that are supported by the API. The data requests (e.g., queries) directed at an endpoint are required to be compliant with various requirements associated with the endpoint. In particular, query statements for an endpoint must be structured to comply with both the syntax of a query language and any requirements which may be stipulated for queries that are supported by an API (or other software interface) for the endpoint.[0041] This process of instructing the LLM to generate a query corresponding to the first data request based on modifying an input prompt to the LLM may proceed iteratively until a successful response is received from the endpoint. [0066] Because GPT-type language models tend to have a large number of parameters, these language models may be considered LLMs. An example GPT-type LLM is GPT-3. GPT-3 is a type of GPT language model that has been trained (in an unsupervised manner) on a large corpus derived from documents available to the public online. [0078] The generated query may then be transmitted to the endpoint 140 via the network 150. For example, a communications module 118 of the code generation engine 114 may be configured to transmit the generated query to an API endpoint associated with a third-party server from which a resource is desired to be retrieved. [0094] In operation 208, the computing system receives, from the LLM, a result including the generated query.); and providing the response to the second prompt in the user interface ([0094] The generated query may be provided to the user device as a response to the first user request. That is, the computing system may output the generated query responsive to receiving the first user request via the user device. [0101] In operation 312, the computing system receives, from the LLM, a result including the generated query. The result may indicate information about the generated query, such as the query language, data fields, arguments, etc. The generated query may be provided to the user device as a response to the first user request. ).
Regarding Claim 2, Almaer et al. teach: The method of claim 1, wherein generating the data characterizing the second prompt further comprises: modifying the received data characterizing the first prompt based on at least one of a type of the artificial intelligence based model, a setting of the artificial intelligence based model, or a configuration for an enterprise in which the artificial intelligence based model is deployed (See rejection of claim 1 specifically [0028] In the present application, the term “generative AI model” may be used to describe a machine learning model. A generative AI model may sometimes be referred to, or may use, a language learning model. [0066] Because GPT-type language models tend to have a large number of parameters, these language models may be considered LLMs. An example GPT-type LLM is GPT-3. GPT-3 is a type of GPT language model that has been trained (in an unsupervised manner) on a large corpus derived from documents available to the public online. [0030] In particular, query statements for an endpoint must be structured to comply with both the syntax of a query language and any requirements which may be stipulated for queries that are supported by an API (or other software interface) for the endpoint. [0041] This process of instructing the LLM to generate a query corresponding to the first data request based on modifying an input prompt to the LLM may proceed iteratively until a successful response is received from the endpoint.).
Regarding Claim 3, Almaer et al. teach: The method of claim 2, wherein the type of the artificial intelligence based model comprises at least one of a foundational model, a multimodal model, a reinforcement learning model, a transfer learning model, or a large language model (LLM) (See rejection of claim 1 and [0028] In the present application, the term “generative AI model” may be used to describe a machine learning model. A generative AI model may sometimes be referred to, or may use, a language learning model. [0066] Because GPT-type language models tend to have a large number of parameters, these language models may be considered LLMs. An example GPT-type LLM is GPT-3. GPT-3 is a type of GPT language model that has been trained (in an unsupervised manner) on a large corpus derived from documents available to the public online.) .
Regarding Claim 4, Almaer et al. teach: The method of claim 2, wherein the setting of the artificial intelligence based model comprises at least one of a temperature, a frequency penalty, a top P- value, or a top K-value (See rejection of claim 2 and [0073] The API call may also include an identification of the language model or LLM to be accessed and/or parameters for adjusting outputs generated by the language model or LLM, such as, for example, one or more of a temperature parameter (which may control the amount of randomness or “creativity” of the generated output) (and/or, more generally some form of random seed as serves to introduce variability or variety into the output of the LLM), a minimum length of the output (e.g., a minimum of 10 tokens) and/or a maximum length of the output (e.g., a maximum of 1000 tokens), a frequency penalty parameter (e.g., a parameter which may lower the likelihood of subsequently outputting a word based on the number of times that word has already been output), a “best of” parameter (e.g., a parameter to control the number of times the model will use to generate output after being instructed to, e.g., produce several outputs based on slightly varied inputs). The prompt generated by the computing system is provided to the language model or LLM and the output (e.g., token sequence) generated by the language model or LLM is communicated back to the computing system.).
Regarding Claim 5, Almaer et al. teach: The method of claim 2, wherein the configuration for the enterprise comprises at least one of language preferences (“best” prompt template), or data masking preferences (See rejection of claim 2, specifically [0028] In some cases, this may include a prompt template. A prompt template may specify that prompts have a certain structure or constrained intents, or that acceptable prompts exclude certain classes of subject matter or intent, such as the production of results or outputs that are violent, pornographic, etc. [0035] The present application discloses improved techniques of generating code for interacting with an endpoint. A system and methods for producing automatically-generated queries using an LLM are described. More particularly, the proposed system is designed to find optimal example(s) of acceptable code for an endpoint that can be used to facilitate query generation. The system may match a user request (e.g., a data retrieval request) to an out of a set of such templates, for an LLM. A prompt template may, for example, comprise a previous query that was accepted by the endpoint or an example of a properly constructed query for the endpoint. The matched template may then be provided in an input prompt to the LLM with instructions to generate a query for submitting to the endpoint.).
Regarding Claim 6, Almaer et al. teach: The method of claim 5, wherein the language preferences comprises tone, cadence, or narrative styles (template may specify that prompts have a certain structure or constrained intents ) (See rejection of claim 5 and [0030] In particular, query statements for an endpoint must be structured to comply with both the syntax of a query language and any requirements which may be stipulated for queries that are supported by an API (or other software interface) for the endpoint.).
Regarding Claim 7, Almaer et al. teach: The method of claim 2, wherein the configuration for the enterprise comprises enterprise specific data (See rejection of claim 2 and [0029] Significant advances have been made in recent years in generative AI models. Different implementations may be trained to create digital art, computer code, conversation text responses, or other types of outputs. Examples of generative AI models include Stable Diffusion by Stability AI Ltd., ChatGPT by OpenAI, DALL-E 2 by OpenAI, and GitHub CoPilot by GitHub and OpenAI. The models are typically trained using a large data set of training data. For instance, in the case of AI for generating images, the training data set may include a database of millions of images tagged with information regarding the contents, style, artist, context, or other data about the image or its manner of creation. The generative AI trained on such a data set is then able to take an input prompt in text form, which may include suggested topics, features, styles or other suggestions, and provide an output image that reflects, at least to some degree, the input prompt. [0066] ChatGPT is built on top of a GPT-type LLM, and has been fine-tuned with training datasets based on text-based chats (e.g., chatbot conversations). ChatGPT is designed for processing natural language, receiving chat-like inputs and generating chat-like outputs.).
Regarding Claim 8, Almaer et al. teach: The method of claim 7, wherein the enterprise specific data comprises at least one of sales expenditure, marketing expenditure, revenue, win rate, statistics, inventory levels, logistics datasets, collections metrics, or lead conversions (See rejection of claim 7 and [0051] For example, an ML model for generating natural language that has been trained generically on publicly-available text corpuses may be, e.g., fine-tuned by further training using the complete works of Shakespeare as training data samples (e.g., where the intended use of the ML model is generating a scene of a play or other textual content in the style of Shakespeare). [0066] ChatGPT is built on top of a GPT-type LLM, and has been fine-tuned with training datasets based on text-based chats (e.g., chatbot conversations). ChatGPT is designed for processing natural language, receiving chat-like inputs and generating chat-like outputs. [0067] A computing system may access a remote language model (e.g., a cloud-based language model), such as ChatGPT or GPT-3, via a software interface (e.g., an application programming interface (API)). Additionally, or alternatively, such a remote language model may be accessed via a network such as, for example, the Internet. In some implementations such as, for example, potentially in the case of a cloud-based language model, a remote language model may be hosted by a computer system as may include a plurality of cooperating (e.g., cooperating via a network) computer systems such as may be in, for example, a distributed arrangement. Notably, a remote language model may employ a plurality of processors (e.g., hardware processors such as, for example, processors of cooperating computer systems). Indeed, processing of inputs by an LLM may be computationally expensive/may involve a large number of operations (e.g., many instructions may be executed/large data structures may be accessed from memory) and providing output in a required timeframe (e.g., real-time or near real-time) may require the use of a plurality of processors/cooperating computing devices as discussed above.).
Regarding Claim 9, Almaer et al. teach: The method of claim 1, wherein the first prompt is provided by the user interface in natural language form (See rejection of claim 1 specifically [0036] When a user provides a first data request (expressed using natural language) for an endpoint, the system is configured to instruct an LLM to generate a query for the endpoint, i.e., by converting the first data request to a corresponding query. [0077] The data request may be expressed in natural language and include information identifying the requested resources. In some implementations, the data request may be received via a user interface on the user device 120.).
Regarding Claim 10, Almaer et al. teach: The method of claim 1, wherein the data corresponding to the response to the second prompt is provided to the user interface in natural language form (See rejection of claim 7 and [0066] GPT-3 has been trained as a generative model, meaning that it can process input text sequences to predictively generate a meaningful output text sequence. ChatGPT is built on top of a GPT-type LLM, and has been fine-tuned with training datasets based on text-based chats (e.g., chatbot conversations). ChatGPT is designed for processing natural language, receiving chat-like inputs and generating chat-like outputs.).
Regarding Claim 14, Almaer et al. teach: The method of claim 1, further comprising: selecting the artificial intelligence based model (an API endpoint associated with a third-party server) based on the second prompt (See rejection of claim 1 specifically, [0030] An endpoint may implement (or expose) a software interface, such as an application programming interface (API), for offering various services to other computer programs. An API contains and is implemented by function calls, which are language statements that request software to perform particular actions and services. The specification of the API describes functions and other parameters that are supported by the API. The data requests (e.g., queries) directed at an endpoint are required to be compliant with various requirements associated with the endpoint. In particular, query statements for an endpoint must be structured to comply with both the syntax of a query language and any requirements which may be stipulated for queries that are supported by an API (or other software interface) for the endpoint. [0041] This process of instructing the LLM to generate a query corresponding to the first data request based on modifying an input prompt to the LLM may proceed iteratively until a successful response is received from the endpoint. [0066] Because GPT-type language models tend to have a large number of parameters, these language models may be considered LLMs. An example GPT-type LLM is GPT-3. GPT-3 is a type of GPT language model that has been trained (in an unsupervised manner) on a large corpus derived from documents available to the public online. [0078] The generated query may then be transmitted to the endpoint 140 via the network 150. For example, a communications module 118 of the code generation engine 114 may be configured to transmit the generated query to an API endpoint associated with a third-party server from which a resource is desired to be retrieved.).
Regarding Claim 17, Almaer et al. teach: The method of claim 1, wherein generating the data characterizing the second prompt further comprises: modifying the received data characterizing the first prompt to restrict a response of the artificial intelligence based model to information contained within a provided dataset object (See rejection of claim 1 and [0028] In some cases, the generative AI model may be implemented with constraints on the acceptable prompts. In some cases, this may include a prompt template. A prompt template may specify that prompts have a certain structure or constrained intents, or that acceptable prompts exclude certain classes of subject matter or intent, such as the production of results or outputs that are violent, pornographic, etc.).
Regarding Claim 19, Almaer et al. teach: The method of claim 1, further comprising: verifying accuracy of the response to the second prompt using an information model communicatively coupled to an enterprise database (correctly formed queries for the endpoint)(See rejection of claim 1 and [0038] More generally, the system identifies an embedding that matches (e.g., nearest neighbor or otherwise closest to) the first embedding, and retrieves a previous query (in the specified query language) that is associated with the identified embedding. The system may, for example, perform a vector search to identify said closest embedding. In at least some implementations, the system may only search embeddings associated with previous queries that are known to have invoked a successful response from the endpoint. That is, the system may identify the closest one of the embeddings associated with correctly formed queries for the endpoint. [0040] The system may then provide the generated query to the endpoint. If the system receives an error response from the endpoint, details about the error, i.e., error data, may be obtained and provided as further input to the LLM. The error data may specify, for example, information on the line number, character number, etc. that contains the erroneous code. The error data is provided as part of an input prompt to the LLM to re-generate the query for the endpoint. Other feedback from the endpoint may be additionally provided to the LLM in the instruction to re-generate the query.).
Regarding Claim 20, Almaer et al. teach: The method of claim 1, wherein the second prompt includes at least one of a dataset object, an indication of outcomes of the dataset, or parameters on how the response should be formulated (See rejection of claim 1 and [0040] The system may then provide the generated query to the endpoint. If the system receives an error response from the endpoint, details about the error, i.e., error data, may be obtained and provided as further input to the LLM. The error data may specify, for example, information on the line number, character number, etc. that contains the erroneous code. The error data is provided as part of an input prompt to the LLM to re-generate the query for the endpoint. Other feedback from the endpoint may be additionally provided to the LLM in the instruction to re-generate the query.).
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.
Claim(s) 11-13, 18, and 21, are rejected under 35 U.S.C. 103 as being unpatentable over Almaer et al. in view of Tsun et al.(US 2024/0296293 A1).
Regarding Claim 11, Almaer et al. teach: The method of claim 1, wherein generating the data characterizing the second prompt further comprises: receiving user historical data characteristics (query matching previous query from user); and generating, based on the historical data characteristics, a blueprint for modifying (previous query is provided, along with the first data request, as input to the LLM ) the first prompt into the second prompt (See rejection of claim 1 and [0035] The system may match a user request (e.g., a data retrieval request) to a “best” prompt template, out of a set of such templates, for an LLM. A prompt template may, for example, comprise a previous query that was accepted by the endpoint or an example of a properly constructed query for the endpoint. The matched template may then be provided in an input prompt to the LLM with instructions to generate a query for submitting to the endpoint. [0036] When a user provides a first data request (expressed using natural language) for an endpoint, the system is configured to instruct an LLM to generate a query for the endpoint, i.e., by converting the first data request to a corresponding query. [0038] More generally, the system identifies an embedding that matches (e.g., nearest neighbor or otherwise closest to) the first embedding, and retrieves a previous query (in the specified query language) that is associated with the identified embedding. [0039] The retrieved previous query is provided, along with the first data request, as input to the LLM. [0041] This process of instructing the LLM to generate a query corresponding to the first data request based on modifying an input prompt to the LLM may proceed iteratively until a successful response is received from the endpoint. [0042] When a user inputs a data request, an embedding associated with the data request may be generated, and compared to embeddings for previous data requests. [0047] Training an ML model refers to a process of learning the values of the parameters (or weights) of the neurons in the layers such that the ML model is able to model the target behavior to a desired degree of accuracy. Training typically requires the use of a training dataset, which is a set of data that is relevant to the target behavior of the ML model. For example, to train an ML model that is intended to model human language (also referred to as a language model), the training dataset may be a collection of text documents, referred to as a text corpus (or simply referred to as a corpus). The corpus may represent a language domain (e.g., a single language), a subject domain (e.g., scientific papers), and/or may encompass another domain or domains, be they larger or smaller than a single language or subject domain.).
Alamer et al. do not specifically teaches underlined limitation: receiving historical user behavior including historical data analysis characteristics; and generating, based on the historical data analysis characteristics, a blueprint for modifying the first prompt into the second prompt.
Tsun et al. teach: receiving historical user behavior including historical data analysis characteristics; and generating, based on the historical data analysis characteristics, a blueprint (a second subset of prompt inputs) for modifying the first prompt into the second prompt ([0044] In some embodiments, prompt generation component 160 determines the messaging intent based on historical activity data of the user of user system 110. For example, prompt generation component 160 determines that the messaging intent is to seek work if the user of user system 110 has recently applied to one or more jobs. [0046] In some embodiments, prompt generation component 160 maps a set of user attributes to a set of one or more prompt inputs using the connection. For example, prompt generation component 160 maps user attributes that are relevant based on the ranking of the connection between the user initiating the electronic messaging and the recipient of the electronic messaging. In some embodiments, prompt generation component 160 maps a shared attribute (e.g., college attended) of attribute data 104 to a prompt input of prompt 106 based on the connection (e.g., the fact that the message sender and message recipient attended the same college). [0048] In some embodiments, input generation component 164 creates an initial prompt using a first subset of prompt inputs of the set of prompt inputs mapped to the user attributes and updating the initial prompt to generate prompt 106 which includes a second subset of prompt inputs of the set of prompt inputs. [0103] In some embodiments, content generation system 100 extracts attribute data from a post based on the selected message intent option 1110. For example, in response to determining that a user is seeking work (e.g., either in response to a selection by the user or an inference by content generation system 100), content generation system 100 extracts attribute data from a post associated with a job that the user is interested in. In some embodiments, content generation system 100 extracts the attribute data from the post based on historical activity data of the user. For example, if the user has recently applied to a job and is now messaging the profile of the person and/or company that posted the job, content generation system 100 can infer that the user intends to talk about that job posting and extracts attribute data from the job posting to use in prompt generation.).
Therefore, it would have been obvious to one of ordinary skilled in the art at the time of the invention was made for Almaer et al. to include the teaching of Tsun et al. above in order to create an initial prompt using a first subset of prompt inputs of a set of prompt inputs mapped to a user attributes data based on historical activity data of the user and updating the initial prompt to generate prompt which includes a second subset of prompt inputs of the set of prompt inputs.
Regarding Claim 12: The method of claim 11, wherein the blueprint is at least partially automatically generated based on metadata (a set of user attributes to a set of one or more prompt inputs) (See Tsun et al. teaching in the rejection of claim 11.).
Regarding Claim 13: The method of claim 1, wherein generating data characterizing the second prompt is based at least on user feedback to historical provided prompt responses (See rejection of claim 11, and Tsun teaching: [0044] In some embodiments, prompt generation component 160 determines the messaging intent based on historical activity data of the user of user system 110. For example, prompt generation component 160 determines that the messaging intent is to seek work if the user of user system 110 has recently applied to one or more jobs. [0048] In some embodiments, input generation component 164 creates an initial prompt using a first subset of prompt inputs of the set of prompt inputs mapped to the user attributes and updating the initial prompt to generate prompt 106 which includes a second subset of prompt inputs of the set of prompt inputs. [0059] Prompt feedback component 168 is a component that receives suggestion 114 from deep learning model 108 and feedback 116 from user system 110 and uses them to generate future prompts. [0061] In some embodiments, prompt feedback component 168 receives feedback from user system 110. For example, user system 110 includes user interface 112 and, as explained with reference to FIGS. 4-9, user interface 112 includes a graphical user interface (such as graphical user interface 400 of FIGS. 4-9). The graphical user interface can include a profile interface displaying suggestion 114 with which a user can interact. For example, the profile interface displays suggestion 114 and the user interacts with the profile interface to refresh the suggestion. In response to receiving this interaction, user system 110 sends feedback 116 to prompt feedback component 168, indicating that the suggestion should be refreshed.).
Therefore, it would have been obvious to one of ordinary skilled in the art at the time of the invention was made for Almaer et al. to include the teaching of Tsun et al. above in order to create an initial prompt using a first subset of prompt inputs of a set of prompt inputs mapped to a user attributes data based on historical activity data of the user and receives suggestion from deep learning model and feedback from user system to generate to updated initial prompt which includes a second subset of prompt inputs of the set of prompt inputs.
Regarding Claim 18: The method of claim 17, wherein the second prompt is configured to reduce hallucinations generated by the artificial intelligence model (See rejection of claim 17 and Tsun et al. teaching, [0087] In some embodiments, prompt 106 includes plan of action 350. Plan of action 350 is a conditioned content generation method to improve relevance, engagement, and diversity of the generated content while mitigating hallucination and prompt injection challenges. For example, instead of or in addition to including step-by-step instructions 310, prompt 106 includes a plan of action 350. In some embodiments, plan of action 350 is an instruction by prompt generation component 160 to deep learning model 108 to generate a plan and ensure the generated plan captures important content requirements of prompt 106. For example, in response to user input indicating that generated content should follow certain content requirements, prompt generation component 160 updates plan of action 350 to include these content requirements.) .
[0061] In some embodiments, prompt feedback component 168 receives feedback from user system 110. For example, user system 110 includes user interface 112 and, as explained with reference to FIGS. 4-9, user interface 112 includes a graphical user interface (such as graphical user interface 400 of FIGS. 4-9). The graphical user interface can include a profile interface displaying suggestion 114 with which a user can interact. For example, the profile interface displays suggestion 114 and the user interacts with the profile interface to refresh the suggestion. In response to receiving this interaction, user system 110 sends feedback 116 to prompt feedback component 168, indicating that the suggestion should be refreshed.
Therefore, it would have been obvious to one of ordinary skilled in the art at the time of the invention was made for Almaer et al. to include the teaching of Tsun et al. above in order to updated initial prompt which includes a second subset of prompt inputs of the set of prompt inputs to improve relevance, engagement, and diversity of the generated content while mitigating hallucination and prompt injection challenges.
Regarding Claim 21: The method of claim 1, further comprising: receiving user feedback characterizing at least one of an acceptance or a rejection of the response to the second prompt; and storing data characterizing the user feedback as training data for improving the artificial intelligence based model, wherein the training data is used for at least one of reinforcement learning, fine tuning, or low-rank adaptation of the artificial intelligence based model (See rejection of claim 1 and [0051] In some embodiments, example generation component 166 uses the identifier to retrieve a suggestion example from a data store (such as data store 240 of FIG. 2). In other embodiments, example generation component 166 generates a suggestion example. For example, example generation component 166 uses a high capacity (e.g., language generation model with many parameters of non-constant values) language generation model to generate a suggestion example. In some embodiments, example generation component 166 stores the suggestion example in a data store (such as data store 240 of FIG. 2). In such embodiments, example generation component 166 may first try to retrieve a suggestion example from a data store and generate the suggestion example if there is no suggestion example available. Further details with regard to example generation component 166 are described below. [0066] In some embodiments, instruction generation component 162 generates an updated set of instructions based on the performance parameter for suggestion 114. For example, prompt generation component 160 generates an initial prompt using an initial set of instructions. Prompt feedback component 168 determines that the initial suggestion generated using the initial prompt has a performance parameter that satisfies the threshold. For example, prompt feedback component 168 uses the inference machine learning model to determine that there are unacceptable contradictions and/or redundancies or prompt feedback component 168 receives negatively labeled feedback 116 from user system 110. Based on the determination by prompt feedback component 168, instruction generation component 162 generates an updated set of instructions. Using the updated set of instructions, prompt generation component 160 generates an updated prompt. Prompt generation component 160 applies deep learning model 108 to the updated prompt to generate an updated suggestion. [0067] In some embodiments, instruction generation component 162 uses a trained prompt generation model to generate the set of instructions. For example, as discussed above, prompt feedback component 168 generates training data using prompts and associated labeled feedback. Prompt generation component 160 trains a prompt generation model using this training data. Instruction generation component 162 uses the trained prompt generation model to generate an updated set of instructions. For example, if a certain set of instructions leads to negative feedback, instruction generation component 162 learns to avoid that set of instructions. Conversely, if a certain set of instructions leads to positive feedback, instructions generation component 162 learns to generate that set of instructions.).
Therefore, it would have been obvious to one of ordinary skilled in the art at the time of the invention was made for Almaer et al. to include the teaching of Tsun et al. above in order prompt feedback component to generate training data using prompts and associated labeled feedback and to train a prompt generation model using this training data to generate an updated set of instructions.
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.
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure. The pertinent art of record Cheng et al. (US 2025/0036670 A1) teach: Large Language Models In Cloud Database Platforms (Users of the cloud database platform can provide queries to instruct one or more LLMs to perform generative natural language processing tasks by manipulating or generating text directly in the cloud database platform with a table valued function. Users can provide input to register or generate one or more LLMs of the cloud database platform for performing the natural language processing tasks. Integrating LLMs into the cloud database platform can improve processing capabilities of the LLMs and save computing resources, as specialized LLMs or application-specific API may no longer be necessary.).
Any inquiry concerning this communication or earlier communications from the examiner should be directed to MOHAMMAD K ISLAM whose telephone number is (571)270-5878. The examiner can normally be reached Monday -Friday, EST (IFP).
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, Paras Shah can be reached at 571-270-1650. 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 K ISLAM/Primary Examiner, Art Unit 2653
Prosecution Insights