DETAILED ACTION
Notice of Pre-AIA or AIA Status
1. The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Continued Examination Under 37 CFR 1.114
2. A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant’s submission filed on 09/29/2026 has been entered.
Response to Arguments/Amendments
3. With respect to 112(a) rejection, the amended claims 1, 9 and 16 overcome the rejection. Thus, the 112(a) rejection has been withdrawn.
With respect to 101 abstract idea rejection, Applicant argues on page 2 of the Remarks that Under Step 2A, Prong One, claims are not properly characterized as merely reciting a mental process because it requires embedding generation, similarity-based retrieval from a stored corpus of utterance-intent pairs, prompt construction using retrieved supervised examples, large-language-model processing and integrated-development-environment execution in an ordered technical workflow.
Response: Step 2A, Prong One: This part of the eligibility analysis evaluates whether the claim recites a judicial exception. As explained in MPEP 2106.04, subsection II, a claim “recites” a judicial exception when the judicial exception is “set forth” or “described” in the claim. Under the broadest reasonable interpretation, steps as recited in claim 1 fall within the mental process groupings of abstract ideas because they cover concepts performed in the human mind, including observation, evaluation, judgment, and opinion. See MPEP 2106.04(a)(2, subsection III. Generating an embedding of the developer-spoken utterance, identifying an utterance-intent pair having closest similarity to the developer-spoken utterance, creating a prompt for a large language model, and incorporating the utterance-intent pair into the prompt are identified as mental processes. Specifically, humans could use pen and paper to calculate the embedding of the developer-spoken utterance, comparing the embedding of the developer-spoken utterance with the embedding of other utterances to identify which an utterance having similarity to the developer-spoken utterance, create a prompt (e.g., instructions), and incorporate the utterance with the intent of the utterance into the prompt. The database, the large language model and the integrated development environment are identified as the additional elements. The additional elements are analyzed under Step 2A, Prong Two.
Under Step 2A, Prong Two, Applicant argues on pages 2-3 of the Remarks that the ordered combination as claimed is not merely the use of a generic computer to identify intent faster. Rather, it defines how the computer system retrieves prior utterance-intent examples, structures those examples as supervised prompt content, and uses the resulting prompt to improve intent detection for voice-driven software-development operations. Accordingly, the claims recite an inventive concept and are patent eligible.
Response: Step 2A, Prong Two: This part of the eligibility analysis evaluates whether the claim as a whole integrates the recited judicial exception into a practical application of the exception. This evaluation is performed by (1) identifying whether there are any additional elements recited in the claim beyond the judicial exception, and (2) evaluating those additional elements individually and in combination to determine whether the claim as a whole integrates the exception into a practical application. See MPEP 2106. 04(d). As discussed above with respect to claim 1, the claim recites the additional elements of the database, the large language model and the integrated development environment. Under BRI, humans could use the book as database to store the utterance-intent pair and search for similar utterances. The large language model and the integrated development environment are used in high level of generality. The LLM is used to generally apply the abstract idea without placing any limits on how the LLM functions. Rather, the limitation only recites the outcome of “generate the intent” and do not include any details about how the “generate” is accomplished. See MPEP 2106.05(f). Claim recites “wherein the intent describes the action, which is to be performed in the integrated development environment”. Claim merely indicates that the intent describes the action, and the action is performed in the integrated development environment. The claimed language does not include any details about how the action is performed in the integrated development environment.
Applicant argues on page 3 of the Remarks that the present claims are analogous in the same legally relevant way with Example 40. In the present claims, the technical problem is accurately identifying developer intent from spoken utterances within an integrated development environment while maintaining responsiveness suitable for voice-driven interaction. The claimed technical mechanism addresses that problem by generating an embedding of the developer-spoken utterances, retrieving a closest matching utterance-intent pair from a database, incorporating the retrieved utterance-intent pair into a prompt as a supervised example, and using the resulting retrieval-augmented prompt to guide a large language model in performing intent detection.
Response: Claim 1 in Example 40 is eligible because it integrates the judicial exception into a practical application by dynamically altering data collection to solve a real technical program (e.g., reducing excess network traffic volume and improving performance.) In the present case, generating an embedding of the developer-spoken utterance, identifying an utterance-intent pair having closest similarity to the developer-spoken utterance, creating a prompt for a large language model, and incorporating the utterance-intent pair into the prompt are identified as mental processes. The claims reciting the large language model provide nothing more than mere instructions to implement an abstract idea on a generic computer. The large language model (LLM) is used to generally apply the abstract idea without placing any limits on how the LLM functions. Rather, these limitations only recite the outcome of “generate the intent” and do not include any details how the “generate” is accomplished. See MPEP 2106.05(f). The recitation of “cause the large language model” in limitation merely indicates a field of use or technological environment in which the judicial exception is performed. Although the additional element “cause the large language model” limits the identified judicial exceptions “generate the intent”, this type of limitation merely confines the use of the abstract idea to a particular technological environment (natural language processing) and thus fails to add an inventive concept to the claim. See MPEP 2106.05 (h). The claim does not integrate the judicial exception into a practical application.
The Applicant’s arguments are not persuasive, and thus for these reasons, Examiner respectfully disagrees.
Claim Rejections - 35 USC § 101
4. 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.
5. Claims 1, 4-9, 12-13, 15-16 and 18-26 are rejected under 35 U.S.C. 101 because the claimed invention is directed to a judicial exception (i.e., a law of nature, a natural phenomenon, or an abstract idea) without significantly more.
All the claims are directed towards the statutory category of a machine/apparatus or process.
Claim 1 recites
“1. (Currently Amended) A system for intent detection comprising:
one or more processors; and
one or more hardware storage devices that store instructions that are executable by the one or more processors to cause the system to:
receive a developer-spoken utterance related to an action of an integrated development environment;
generate an embedding of the developer-spoken utterance;
identify an utterance-intent pair having a closest similarity to the developer-spoken utterance as a few-shot example by searching a database using the embedding of the developer-spoken utterance, the database including utterance-intent pairs;
create a prompt for a large language model to identify an intent associated with the developer-spoken utterance;
incorporate the utterance-intent pair into the prompt as a supervised example of an utterance and a corresponding intent for an intent-detection task performed by the large language model, wherein the prompt includes the few-shot example and the developer-spoken utterance, and wherein the intent describes the action, which is to be performed in the integrated development environment;
cause the large language model to use the prompt to generate the intent, wherein the large language model is tasked, by the intent-detection task, with identifying the intent within a predetermined timing requirement for viability within the integrated development environment, and wherein the intent is processed by the integrated development environment; and
process the intent in the integrated development environment.”
Claims 1, 9 and 16 recite substantially the same concept but do so in the context of a system, a method and a device.
The last limitations of Claims 1, 9 and 16 are slightly different.
Claim 1 recites “process the intent in the integrated development environment.”
Claim 9 recites “transmitting the intent to the integrated development environment.
Claim 16 recites “apply the intent in the integrated development environment.”
The limitations in independent claims as drafted cover amental processes. More specifically, the underlaying abstract idea revolved around what happens once a human listen to an utterance, obtains example similarity with the utterance, identifies an intent of the utterance from the intent of the example, performs an action corresponding with the intent or send the intent to somebody else or apply the intent.
The judicial exception is not integrated into a practical application. In particular, claims recite additional elements of “one or more processes”, “a memory”, “one or more hardware devices”. The additional element(s) or combination of elements such as one or more processes, a memory, one or more hardware devices, in the claim(s) other than the abstract idea per se amount(s) to no more than (i) mere instructions to implement the idea on a computer, and/or (ii) recitation of generic computer structure that serves to perform generic computer functions that are well-understood, routine, and conventional activities previously known to the pertinent industry. Viewed as a whole, these additional claim element(s) do not provide meaningful limitation(s) to transform the abstract idea into a patent eligible application of the abstract idea such that the claim(s) amounts to significantly more than the abstract idea itself. Therefore, the claim(s) are rejected under 35 U.S.C. 101 as being directed to non-statutory subject matter. There is no further improvement to the computing device other than identifying an intent of the user’s utterance and utilizing the intent in some situations. The mere recitation of “one or more processes”, “a memory”, “one or more hardware devices” and/or the like is akin of adding the word “apply it” and/or “use it” with a computer in conjunction with the abstract idea. Paragraph
[0068] discloses “The identification of the intent has to be performed within tight timing requirements in order to be viable in a target system. For at least these reasons, the operations used to generate the prompt need to be performed on a computing device. The operations performed are inherently digital. A human mind cannot interface directly with a CPU, or network interface card, or other processor, or with RAM or digital storage, to read and write the necessary data and perform the necessary operations and processing steps taught herein.” Paragraph [0090] discloses “One or more hardware storage devices is disclosed having stored thereon computer- executable instructions that are structured to be executable by one or more processors of a computing device to thereby cause the computing device to perform actions that: obtain through a voice user interface a developer-spoken utterance to perform an action in the integrated development environment; create a prompt for a large language model to predict an intent of the developer-spoken utterance, wherein the prompt includes few-shot examples and the developer-spoken utterance, wherein the few-shot examples are supervised data that train the large language model on an intent detection task; apply the prompt to the large language model; obtain an intent corresponding to the developer-spoken utterance from the large language model; and apply the intent in the integrated development environment.”
As filed in the specification, the computer is listed as a general-purpose computer and is mainly used as an application thereof. Accordingly, these 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.
The Claims reciting the large language model provide nothing more than mere instructions to implement an abstract idea on a generic computer. The large language model (LLM) is used to generally apply the abstract idea without placing any limits on how the LLM functions. Rather, these limitations only recite the outcome of “generate the intent” and do not include any details how the “generate” is accomplished. See MPEP 2106.05(f). The recitation of “cause the large language model” in limitation merely indicates a field of use or technological environment in which the judicial exception is performed. Although the additional element “cause the large language model” limits the identified judicial exceptions “generate the intent”, this type of limitation merely confines the use of the abstract idea to a particular technological environment (natural language processing) and thus fails to add an inventive concept to the claim. See MPEP 2106.05 (h).
Claim 9 recites transmitting the prompt to the large language model and transmitting the intent to the integrated development environment. For precedential case law-refer to MPEP 2106.05(d)(II), 2106.05(f), and 2106.05(g), for example. For examples of useful cases include: Receiving or transmitting data over a network, e.g., using the Internet to gather data, Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362; Claims that amount to nothing more than an instruction to apply the abstract idea using a generic computer do not render an abstract idea eligible. Alice Corp., 573 U.S. at 223, 110 USPQ2d at 1983; Use of a computer or other machinery in its ordinary capacity for economic or other tasks. Affinity Labs v. DirecTV, 838 F.3d 1253, 1262, 120 USPQ2d 1201, 1207 (Fed. Cir. 2016) (cellular telephone); TLI Communications LLC v. AV Auto, LLC, 823 F.3d 607, 613, 118 USPQ2d 1744, 1748 (Fed. Cir. 2016) (computer server and telephone unit); Selecting information, based on types of information and availability of information in a power-grid environment, for collection, analysis and display, Electric Power Group, LLC v. Alstom S.A. 830 F.3d 1350, 1354-55, 119 USPQ2d 1739, 1742 (Fed. Cir. 2016).
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 the integration of the abstract idea into a practical application, the additional element of using a computer is noted as a general computer. Mere instructions to apply an exception using a generic computer component cannot provide an inventive concept. The claims are not patent eligible.
The dependent claims do not remedy the issues noted above. More specifically, claims 4 and 12 recite using the cosine similarities to find similar utterances. This reads on the human could use cosine formula to find similar utterances. Claims 5, 13, 18 recites the first set of instruction in the prompt. No additional limitations are presented. Claims 6 and 19 recite the second set of instruction in the prompt. No additional limitations are presented. Claims 7-8, 15, 20 define the large language model. No additional limitations are presented. Claim 21 recites mental processes of listening to the utterance and write down the transcription of the utterance on the paper. No additional limitations are presented. Claim 22 merely describes what the intent is. No additional limitations are presented. Claim 23 describes the format of the supervised example. No additional limitations are presented. Claim 24 indicates the large language model resides on an external server, and wherein the system communicates with the large language model through one or more application programming interfaces over a network. It can be viewed as nothing more than an attempt to generally link the use of the judicial exception to the technological environment of computers. Claim 25 recites using a neural transformer with attention to generate the embedding of the developer-spoken utterance. There are no details about the neural transformer model with attention or how the neural transformer model with attention operates to derive the embedding of the developer-spoken utterance. The neural transformer model with attention is used to apply the abstract idea (e.g., perform the mental process of calculation) using the equation. Claim 26 recites mental processes of selecting the utterance-intent pair.
For at least the supra provided reasons, claims 1, 4-9, 12-13, 15-16 and 18-26 are rejected under 35 U.S.C. 101 as being directed to non-statutory subject matter.
Allowable Subject Matter
6. Claims 1, 4-9, 12-13, 15-16 and 18-26 are allowed in view of the prior art of record. The claims stand rejected under 101 Abstract idea, and for the application to pass to allowance this rejection needs to be overcome. Any amendments to overcome the 101 rejection that results in any change in scope require further search and/or consideration to determine its allowability.
The following is a statement of reasons for the indication of allowable subject matter: the prior art(s) taken alone or in combination fail(s) to teach the following element(s) in combination with the other recited elements in the claim(s).
“incorporate the utterance-intent pair into the prompt as a supervised example of an utterance and a corresponding intent for an intent-detection task performed by the large language model, wherein the prompt includes the few-shot example and the developer-spoken utterance, and wherein the intent describes the action, which is to be performed in the integrated development environment;
cause the large language model to use the prompt to generate the intent, wherein the large language model is tasked, by the intent-detection task, with identifying the intent within a predetermined timing requirement for viability within the integrated development environment, and wherein the intent is processed by the integrated development environment; and” as recited in claim 1.
Claims 9 and 16 recite similar features as claim 1.
The closest prior arts found as follows.
The closest prior arts found as follows.
Dinu et al. (US 2024/0354319 A1.) In this reference, Dinu et al. disclose generating a few-shot examples that includes the similar example user input, inputting the few-shot examples into the large language model, determining the intent of the user input, and output the response to the user input [0042] using a few-shot prompt that includes the most similar example user inputs, the corresponding canonical form inputs, and the current conversation (e.g., the dialog history) with the user, as discussed in greater detail herein in conjunction with at least FIGS. 6-7. In such cases, the canonical form generator 204 can determine the number of most similar example user inputs by generating an embedding (e.g., a vector embedding) of the user input in a semantic or latent space, such as by inputting the user input into a sentence transformer or other trained machine learning model that outputs the embedding as a vector, and then comparing the embedding of the user input to embeddings of the example user inputs in the canonical form input definitions 212 to determine one or more similar example user inputs whose embeddings are closest, according to a distance metric, to the embedding of the user in the semantic or latent space. Any technically feasible distance metric can be used in at least one embodiment, including well-known distance metrics. The canonical form generator 204 then generates a few-shot prompt that includes the similar example user inputs, corresponding predefined canonical form inputs, and/or the current conversation, with the few-shot prompt being in the syntax of the formal modeling language, [0035] the language model 220 can be a large language model (LLM), [0042] prompting the language model 220 to generate the canonical form input using a few-shot prompt…the canonical form generator 204 inputs the few-shot prompt into the language model 220, which outputs a canonical form input. In at least one other embodiment, the canonical form generator 204 can input the user input 202 (and optionally context, such a history of the current conversation) into a trained machine learning model, such as a p-tuned LLM, that was trained to output a canonical form in response to input text, e.g., to translate the input text into a canonical form. In such cases, the trained machine learning model can output a canonical form input when the canonical form generator 204 inputs the user input 202 (and optional context) into the trained machine learning model, [0037] a “canonical form” can include a constrained semantic representation (e.g., summarization, short description, intent definition).) Dinu et al. generates the prompt and causes the large language model to use the prompt to determine the intent of the user’s input. However, Dinu et al. does not teach and/or suggest creating a prompt for a large language model and incorporating the utterance-intent pair into the prompt as a supervised example of an utterance and causing the large language model to use the prompt to generate the intent. Thus, Dinu et al. fail to teach and/or suggest the allowable subject matter.
Shabat et al. (US 2024/0203404 A1.) In this reference, Shabat et al. disclose using the LLM to identify intent of the spoken utterance (Shabat et al. [0028] The pre-trained text encoder 107 may be a previously trained ML model or a combination of various previously trained ML models that can be fine-tuned. For example, the pre-trained text encoder 107 may itself correspond to a LLM, such as LaMDA, BERT, T5, Meena, GPT-3, and/or another previously trained LLM. Notably, these previously trained LLMs have been previously trained on enormous amounts of diverse data. These LLMs have a plurality of ML layers and hundreds of millions to hundreds of billions of ML parameters. For example, textual data may be provided as input across these previously trained LLMs to generate LLM output, such as a probability distribution over intents, and one or more intents present in the textual data may be determined based on the probability distribution over the intents, [0038] Accordingly, the fine-tuned LLM 160 can, based on processing the textual data with the text encoder 152 and the audio encoder 153, provide one or more semantic representations corresponding to the spoken utterance in the audio data 150. For instance, as depicted in FIG. 1B, the fine-tuned LLM 160 can generate predicted intent(s) 170 (or simply intent(s) 170) of the spoken utterance in the audio data 140 and one or more corresponding slot values for one or more parameters associated with each of the one or more predicted intents.) Shabat et al. uses the previously trained LLMs to generate a probability distribution over intents and one or more intents present in the textual data is determined based on the probability distribution over to the intent. However, Shabat et al. does not teach and/or suggest creating a prompt for a large language model and incorporating the utterance-intent pair into the prompt as a supervised example of an utterance and causing the large language model to use the prompt to generate the intent. Thus, Shabat et al. fail to teach and/or suggest the allowable subject matter.
Du et al. (US 2025/0298971 A1.) In this reference, Du et al. disclose converting speech to text (i.e., generating a transcription), using LLM to analyze an intention associated with a corresponding portion of the transcription (Du et al. [0155] An intent extractor 920 may be trained and configured to analyze an intention(s) associated with a corresponding portion of the transcription stream 130 at a given point in time. For example, the intent extractor 920 may be trained as a classifier to classify a current intention of the speaker 100 from among a known plurality of intentions, including, e.g., to see, to purchase, to show, to give/receive, to visit, to help, and so on. The intent extractor 920 may be obtained by fine-tuning a pre-trained LLM for intent detection, [0187] The transcript may be augmented with entity, emotion, and/or intent information (1206). Additionally, or alternatively, the summary may similarly be augmented with entity, emotion, and/or intent information (1208). As described above, the transcription and summarizing may occur concurrently with the augmentation(s), in which case the augmentation may or may not be displayed. In other examples, the augmentations may be added at a later time, e.g., to selected, buffered transcriptions/summaries, in which case the transcription/summary may initially be provided without augmentation. That is, the transcription and/or the summary may be displayed without or without inclusion of the entity, emotion, and/or intent information, depending, e.g., on user preferences, and the entity, emotion, and/or intent information may be added at a later time, e.g., to content within the transcription buffer 128 or the summary buffer 132.) Du et al. fine-tunes the pre-trained LLM to detect the intention of the transcription at a given point in time. However, Du et al. does not teach and/or suggest creating a prompt for a large language model and incorporating the utterance-intent pair into the prompt as a supervised example of an utterance and causing the large language model to use the prompt to generate the intent. Thus, Du et al. fail to teach and/or suggest the allowable subject matter.
Conclusion
7. The prior art made of record and not relied upon is considered pertinent to application’s disclosure. See PTO-892.
a. Chen et al. (US 2024/0274286 A1.) In this reference, Chen et al. discloses a method and a system for training a large language model using few-shot learning.
b. Nagaraju et al. (US 2024/0185001 A1.) In this reference, Nagaraju et al. discloses a method and a system for training a large language model by changing the few-shot examples.
c. Shaikh et al. (US 2024/0303439 A1.) In this reference, Shaikh et al. discloses a method and a system for using a large language model to extract linguistic information with relatively small quantity of training examples.
8. Any inquiry concerning this communication or earlier communications from the examiner should be directed to THUYKHANH LE whose telephone number is (571)272-6429. The examiner can normally be reached Mon-Fri: 9am-5pm.
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, Andrew C. Flanders can be reached on 571-272-7516. 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.
/THUYKHANH LE/Primary Examiner, Art Unit 2655
