DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Claim Interpretation
As per Claim 5:
“it” in line 2 of claim 5 is interpreted as referring to “the structured representation” (not “the declared schema”).
As per Claim 12:
“incrementally amend the partial continuation before completion” is interpreted as a function of “intermediate results” (not where “incrementally amend the partial continuation before completion” is an addition step for the method of claim 1 that is not necessarily related to the intermediate results)
As per Claim 30:
“LLM” in claim 30 is interpreted as an acronym for “large language model” (see claims 1 and 29 which explicitly define “LLM” as an acronym for “large language model”).
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 3-9, 11, 13, 15-16, 18-19, 21-27, and 28, 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 original Specification (i.e. the original Specification of Parent Application 18/301,615, hereafter original Specification, where this application is a continuation, and not a continuation-in-part) does not have written description for the following:
It is not clear where the original Specification supports the limitations of claims 3-9, 11, 13, 15-16, 18-19, 21-27, and 28 (for claim 28, especially the “programming assistant”, “content creation”, “data analytics”, “legal research”, “education”, “travel”, and “security” applications/services).
The dependent claims include the issues of their respective parent claims.
Claims 5, 14, 20, 21, 23, 27, and 30, are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
As per Claim 5:
“the declared schema” in line 2 of claim 5 lacks antecedent basis (claim 2 first introduces “a declared schema” but claim 5 depends on claim 1).
As per Claim 14:
“the continuation” in line 2 of claim 14 is ambiguous (it can refer to either “a partial continuation” in line 5 of claim 1, or to “a revised continuation” in the last line of claim 1).
As per Claim 20:
“the… context” in “the prompt or context” (where “the prompt or context” is interpreted as “the prompt or [the] context”) lacks antecedent basis.
As per Claim 21:
“the inference task” in line 2 of claim 21 lacks antecedent basis.
As per Claim 23:
“the machine readable language” in line 2 of claim 23 lacks antecedent basis.
As per Claim 27:
“the continuation” in line 1 of claim 27 is ambiguous (it can refer to either “a partial continuation” in line 5 of claim 1, or to “a revised continuation” in the last line of claim 1).
“the validation or policy outcome” in line 2 of claim 27 lacks antecedent basis.
As per Claim 30:
“the output or response of the LLM-based system to that prompt” in the last line of claim 30 lacks antecedent basis.
Allowable Subject Matter
Claims 1-2, 10, 12, 17, and 29, are allowed.
Claim 30 would be allowable if rewritten or amended to overcome the rejection(s) under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), 2nd paragraph, set forth in this Office action.
Claims 14 and 20 would be allowable if rewritten to overcome the rejection(s) under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), 2nd paragraph, set forth in this Office action and to include all of the limitations of the base claim and any intervening claims.
The following is a statement of reasons for the indication of allowable subject matter:
As per Claim(s) 1 (and similarly claim[s] 29, and consequently claim[s] 2-28 which depend on claim[s] 1), the prior art of record does not teach or suggest the combination of all limitations in claim(s) 1, including (i.e. in combination with the remaining limitations in claim[s] 1) A computer implemented method of improving the accuracy or reliability of an AI system including a large language model (LLM) based system, in which the LLM-based system uses a deep learning model capable of processing natural language and the AI system is capable of generating a sequence of reasoning steps; the method comprising: (i) generating, by the LLM-based system, a partial continuation for a prompt; (ii) during generation of the partial continuation, providing the partial continuation to a non LLM processing engine as a structured, machine readable representation that is distinct from natural language text; and (iii) amending the partial continuation before completion in response to output from the non LLM processing engine, and producing a revised continuation for the prompt.
As per Claim(s) 30, the prior art of record does not teach or suggest the combination of all limitations in claim(s) 30, including (i.e. in combination with the remaining limitations in claim[s] 30) A computer implemented AI based application with improved accuracy or reliability, comprising: a user interface configured to receive a prompt and to present a response; and in which the AI-based application is configured to exchange data with an LLM based system coupled to a non LLM processing engine, in which, during generation of a partial continuation for the prompt, the LLM based system produces a structured machine readable representation that conforms to a declared schema and is distinct from natural language text, and the non LLM processing engine processes the structured representation to amend the partial continuation before completion to produce a revised continuation for the prompt; and the AI-based application is configured to enable an end-user to provide the prompt to the LLM-based system using the user interface and to display the revised continuation as the output or response of the LLM-based system to that prompt.
For prior art evaluation purposes, “a structured, machine readable representation that is distinct from natural language text” (in claims 1 and 29) and “a structured machine readable representation that… is distinct from natural language text” (in claim 30) are interpreted as where the structured, machine readable representation is not “natural language text” (since a structured, machine readable representation cannot be fairly interpreted as being “distinct from natural language text” if it is natural language text). An alternative interpretation was considered where “distinct from natural language text” could be interpreted as where the structured machine readable representation differs from a particular piece of natural language text (e.g., the structured machine readable representation could be other natural language text that is not the letter “a”, or the word “word”, or any other specific example of natural language text) but the examiner declined to apply this alternative interpretation because anything that is natural language text cannot fairly be considered to have the characteristic of being “distinct from natural language text” when the natural language text is, itself, natural language text.
12153874 (LATE filing date) teaches “In another example, the text rewriting model 134 can be configured to rewrite an entirety of the output generated by the generative model 120 (instead of a sentence or phrase corresponding to a citation), where the text rewriting model 134 rewrites the text using the entirety of the prompt employed by the generative model 120 to generate the output. A disadvantage of such approach is that the text rewriting model 134 is unable to rewrite the text generated by the generative model 120 until the generative model 120 has generated the entirety of the output. In contrast, the approaches described above allow for the text rewriting model 134 to rewrite portions of an output (e.g., a phrase or sentence) generated by the generative model 120 prior to the generative model 120 completing the output. Another disadvantage of configuring the text rewriting model 134 to rewrite an entirety of the output of the generative model 120 is that the text rewriting model 134 may unnecessarily rewrite a portion of the output, thereby consuming additional processing resources. In contrast, the communications diagram 200 indicates that text is given to the text rewriting model 134 only when the classifier 130 has predicted that the text is not supported by content of a source (e.g., webpage) identified in a citation for the text. Moreover, providing the text rewriting model 134 with a relatively small amount of text (e.g., a sentence rather than several sentences) results in faster end to end rewriting time, as more time (and processing resources) is needed to build a query-key-value cache for longer input sequences, and each new generation time step will require more resources since the text rewriting model 134 is attending to a longer context length (when the text rewriting model 134 is a transformer language model)” (col. 9, lines 4-34) and “With reference to FIG. 7, the GUI 300 is shown, where the output field 304 includes additional text generated by the generative model 120. FIG. 7 is set forth to illustrate that the text rewriting model 134 can rewrite one portion of the output generated by the generative model 120 while the generative model 120 continues to generate output. That is, the text rewriting model 134 can operate in parallel with the generative model 120” (col. 10, lines 55-62). This reference does not qualify as prior art.
2024/0403290 (LATE filing date) teaches “As noted above, LLMs may provide incomplete or inaccurate responses if the computer language and/or logic of the prompt and/or response is too complex. This may be due in part to the complexity of the grammar required by the corresponding computer language, a required syntax (e.g., spelling, capitalization) of the computer language, or some other challenge related to a format of the computer language. For example, highly nested logic structures (e.g., as in JSON) can cause the LLM to respond in an incomplete and/or inaccurately formatted response. Even if the LLM is instructed to output in a custom computer language that is optimized for LLM responses and parsability, the LLM may still, at time, provide outputs with errors, formatting inconsistencies, and/or the like. Accordingly, it can be advantageous for the system to identify mistakes (e.g., incompleteness, inaccuracies, and/or the like) in responses from the LLM that are in a computer language, such as a custom computer language. These mistakes may be quantified, categorized, and/or otherwise tracked to further improve a syntax, format, terminology, and/or grammatical structure of the custom computer language. The system can recursively and/or automatically track these mistakes and make updates to the format of the custom computer language based on the mistakes, and with the object off minimizing the mistakes and further optimizing the custom computer language. A user may support in the analysis, tracking, and/or updating of the format of the custom computer language. The system may track progress over time in the updates to the grammar to help ensure that the updates to the grammar are effective at reducing mistakes in future LLM responses. For example, the system may compare a number of mistakes in a first response with a number of mistakes in a second response to a prompt with the updated grammar” (paragraph 69). This reference does not qualify as prior art.
2024/0320444 (LATE filing date) teaches “In at least some implementations, the content generation engine 114 may iteratively perform the steps of receiving user selections of desired portions from outputs of the generative AI model, deriving modified input prompts based on the user selections, and providing the modified prompts to the generative AI model 112” (paragraph 81). This reference does not qualify as prior art.
Double Patenting
For clarity of the record, NO Double Patenting rejections are required between the claims of this application and the claims of the Parent/Sibling applications because the claims of the Parent/Sibling application do not teach or suggest (i) generating, by the LLM-based system, a partial continuation for a prompt; (ii) during generation of the partial continuation, providing the partial continuation to a non LLM processing engine as a structured, machine readable representation that is distinct from natural language text; and (iii) amending the partial continuation before completion in response to output from the non LLM processing engine, and producing a revised continuation for the prompt (for claims 1 and 29) and during generation of a partial continuation for the prompt, the LLM based system produces a structured machine readable representation that conforms to a declared schema and is distinct from natural language text, and the non LLM processing engine processes the structured representation to amend the partial continuation before completion to produce a revised continuation for the prompt; (for claim 30) (Sibling Patent US 11,989,527 claims where input data provided to an LLM so that the LLM generates improved first output is a correction and/or a change of continuation text output generated by the LLM [claim 2] and where the first output from the LLM is a partial continuation made before the LLM has stopped generating or while the LLM is still generating [claim 19] but claims 2 and 19 both depend on claim 1, and so single claim [incorporating the limitations of its parent claims] includes all of the limitations of claims 1, 29, and 30; Sibling Patent 12,430,503 claims where the non-LLM system analyzes an LLM-based system output for accuracy and provides corrections so that the LLM-based system generates a new continuation using the corrections [claim 1] and where the input from the LLM is a partial continuation made before the LLM has stopped generating or while the LLM is still generating [claim 8], but does not specifically claim where the partial continuation is amended before completion [as opposed to where the partial continuation is generated before completion and is amended after completion])
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to ERIC YEN whose telephone number is (571)272-4249. The examiner can normally be reached M-F 12:00PM -8:30PM EST.
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, RICHEMOND DORVIL can be reached at (571)272-7602. 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.
EY 6/19/2026
/ERIC YEN/ Primary Examiner, Art Unit 2658
Prosecution Insights