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 .
Status
This instant application No. 18/444,094 has claims 1-20 pending.
Priority
There is no priority being claimed. The effective filing date for this instant application is February 16, 2024.
Drawings
The drawings filed on February 16, 2024 are acceptable for examination purposes.
Abstract
The abstract of the disclosure is objected due to the use of implied language. Note that in the abstract, the language should be clear and concise and should not repeat information given in the title. It should avoid using phrases which can be implied, such as, “The disclosure concerns,” “The disclosure defined by this invention,” “The disclosure describes,” etc… See MPEP § 608.01(b).
Note that in the abstract, Applicant cites “A…system is described” on lines 1-2. This citation clearly provokes the use of implied language. Correction is required (e.g., removal of the entire first sentence of the abstract).
Information Disclosure Statement
As required by M.P.E.P. 609(C), the Applicant’s submission of the Information Disclosure Statement filed on May 30, 2024 is acknowledged by the Examiner and the cited references have been considered in the examination of the claims. As required by M.P.E.P. 609 C(2), a copy of the PTOL-1449 initialed and dated by the Examiner is attached to the instant Office action.
Claim Objections
Claims 2, 3, 13, 16, and 17 are objected for citing an indefinite term “can be” in each claim since implementation of any elements following such term is optional (e.g., can or cannot). Revision and/or correction are required.
Claim 11 is objected for having grammatical error in the citation of the one or more…models includes a first…model. It is believed the one or more…models include a first…model is more appropriate. Correction is required.
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.
The claimed invention in claims 1-20 are directed to a judicial exception (i.e., an abstract idea) without significantly more.
Claims 1-20 pass step 1 of the 35 U.S.C. 101 analysis since each claim is either directed to a method, a computing system comprising one or more processors and one or more storage media storing instruction executed by the one or more processors (i.e., hardware components per [0173] of instant specification).
Claims 1, and 15 recite each, in part, steps that are directed to an abstract idea (“Courts have examined claims that required the use of a computer and still found that the underlying, patent-ineligible invention could be performed via pen and paper or in a person’s mind.” Versata Dev. Group v. SAP Am., Inc., 793 F.3d 1306, 1335, 115 USPQ2d 1681, 1702 (Fed. Cir. 2015)). Each claim recites the limitations of storing…data associated with at least one synthetic query…; determining a subset…that are relevant to a particular user query… The limitations, as drafted, are parts of a process that, under its broadest reasonable interpretation, covers performance of the limitations in the mind but for the recitation of generic computer components (e.g., writing down, on paper, data associated with a synthetic query generated by a machine-learning model; and mentally determining a subset of the computing tools that are relevant to a user query based on the written data on paper). That is, other than reciting generic components (e.g., processor, memory, and computer-executable instructions), nothing in the claim precludes the limitations from being performed in the human mind per step 2A – prong 1 of the Abstract Idea Analysis. Thus, the limitations are parts of a mental process. Further, the claims recite additional steps of generating at least one prompt for at least one machine-learned sequence processing models…; and generating a response to the particular user query based at least in part on an output of the at least one machine-learned sequence processing model… which are extra-solution activities (per step 2A – prong 2 of the Abstract Idea Analysis) that cannot be integrated into a practical application (e.g., the elements recite trivial elements that occurred or would occur after the mental process) since both the prompt and output generations are often referred to as tool-use or function-calling without specifying a non-conventional machine or architecture that operates in a new way to overcome a technological problem or providing a technical improvement to the computer’s operation (e.g., speed, memory usage, processing capacity) in a non-generic sense. Each of the additional limitation(s) is no more than mere instructions to apply the exception using a generic computer component (e.g., processor, memory, and computer-executable instructions). The extra-solution activity in step 2A - prong 2 are reevaluated in step 2B to determining if each limitation is more than what is well-understood, routine, conventional activity in the field. The background of the limitations does not provide any indication that the computer components (e.g., processor, memory, and computer-executable instructions) are not off-the-shelf computer components. The Symantec, TLI, and OOP Techs court decisions cited in MPEP 2106.05(d)(II) indicate that mere receiving, generating, storing, determining, identifying, and transmitting of data over a network are a well-understood, routine, and conventional functions when claimed in a merely generic manner (as it is here). Accordingly, a conclusion that the claims are well-understood, routine, conventional (WURC) activity is supported under Berkheimer Option 2. For these reasons, there is no inventive concept in each claim, thus, the claims are ineligible.
Claims 2 and 16 recite in each claim additional elements of providing…at least one synthetic query generation request…; and obtaining…at least one synthetic query… which are extra-solution and WURC activities similar to the above analysis. These steps merely describe the functional mechanism of generating the data (synthetic queries) and do not add any technical improvement to the computing system’s structure of operation. Further, the steps are conventional API/computer interactions when requesting generative AI output. Thus, the claims are ineligible.
Claims 3 and 17 recite in each claim an additional element of providing at least one prompt to the…models which is an extra-solution and WURC activity similar to the above analysis. This step merely specifies the use of prompt and including tool documentation as input to the machine-learning model to achieve a desired informational output which does not add any technical improvement to the computing system’s structure of operation. Further, this step is a conventional API/computer interactions of inputting prompts when requesting generative AI output. Thus, the claims are ineligible.
Claims 4 and 18 recite in each claim additional elements of providing…a plurality of synthetic query generation requests…; and varying a temperature… which are extra-solution and WURC activities similar to the above analysis. These steps merely describe inputting requests into a processing model and applying known hyperparameter tuning technique to generate output data based on the inputted requests. The steps do not add any technical improvement to the computing system’s structure of operation and are conventional API/computer interactions when requesting generative AI output. Thus, the claims are ineligible.
Claims 5 and 19 recite in each claim an additional element of storing the tool documentation… which can be implemented in a human mind and/or with the aid of pen/paper similar to the above analysis (e.g., writing down the tool documentation on paper in response to output received from the processing model). Thus, the claim is ineligible.
Claims 6 and 20 recite in each claim additional elements of encoding…the at least one synthetic query…; and …storing at least one embedding of the at least one synthetic query which can be implemented in a human mind and/or with the aid of pen/paper similar to the above analysis (e.g., mentally encoding the query as binary vector (e.g., 10101100) and writing down the vector representation of the query on paper). Thus, the claim is ineligible.
Claim 7 merely provides a definition for the subset of…computing tools includes less than all…computing tools. Thus, the claim is ineligible.
Claim 8 merely provides definitions for the one or more machine-learned…models and the at least one machine-learned…model to include a/the first…model. Thus, the claim is ineligible.
Claim 9 recites an additional element of performing at least one sparse similarity-based retrieval method to compare the particular user query with the data… which can be implemented in a human mind and/or with the aid of pen/paper similar to the above analysis (e.g., mentally performing the sparse similarity-based comparison of the query with the data and writing down the result on paper). Thus, the claim is ineligible.
Claim 10 recites an additional element of performing at least one dense similarity-based retrieval method to compare the particular user query with the data… which can be implemented in a human mind and/or with the aid of pen/paper similar to the above analysis (e.g., mentally performing the dense similarity-based comparison of the query with the data and writing down the result on paper). Thus, the claim is ineligible.
Claim 11 merely provides a definition for the…processing models includes a first large language model. Thus, the claim is ineligible.
Claim 12 merely provides a definition for the at least one…processing model includes a first large language model. Thus, the claim is ineligible.
Claim 13 recites, in part, steps that are directed to an abstract idea (“Courts have examined claims that required the use of a computer and still found that the underlying, patent-ineligible invention could be performed via pen and paper or in a person’s mind.” Versata Dev. Group v. SAP Am., Inc., 793 F.3d 1306, 1335, 115 USPQ2d 1681, 1702 (Fed. Cir. 2015)). The claim recites the limitations of storing…data associated with the at least one synthetic query; processing…a particular user query…based at least in part on the data…for each of the…computing tools. The limitations, as drafted, are parts of a process that, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components (e.g., writing down results of the synthetic query; and mentally processing/comparing a user query using results associated with the synthetic query for each computing tool). That is, other than reciting generic components (e.g., processor, memory, and computer-executable instructions), nothing in the claim precludes the limitations from being performed in the human mind per step 2A – prong 1 of the Abstract Idea Analysis. Thus, the limitations are parts of a mental process. Further, the claim recites additional steps of providing…at least one generation request…; obtaining…at least one synthetic query… which are extra-solution activities (per step 2A – prong 2 of the Abstract Idea Analysis) that cannot be integrated into a practical application (e.g., the elements recite trivial elements that occurred or would occur after the mental process) since both providing and obtaining are generic API interactions of utilizing a machine-learning model often referred to as tool-use or function-calling without specifying a non-conventional machine or architecture that operates in a new way to overcome a technological problem or providing a technical improvement to the computer’s operation (e.g., speed, memory usage, processing capacity) in a non-generic sense. Each of the additional limitation(s) is no more than mere instructions to apply the exception using a generic computer component (e.g., processor, memory, and computer-executable instructions). The extra-solution activity in step 2A - prong 2 are reevaluated in step 2B to determining if each limitation is more than what is well-understood, routine, conventional activity in the field. The background of the limitations does not provide any indication that the computer components (e.g., processor, memory, and computer-executable instructions) are not off-the-shelf computer components. The Symantec, TLI, and OOP Techs court decisions cited in MPEP 2106.05(d)(II) indicate that mere receiving, generating, storing, determining, identifying, and transmitting of data over a network are a well-understood, routine, and conventional functions when claimed in a merely generic manner (as it is here). Accordingly, a conclusion that the claims are well-understood, routine, conventional (WURC) activity is supported under Berkheimer Option 2. For these reasons, there is no inventive concept in each claim, thus, the claims are ineligible.
Claim 14 recites an additional element of determining…a subset of the…tools that are relevant… which can be implemented in a human mind and/or with the aid of pen/paper similar to the above analysis (e.g., mentally determining a relevant subset of tools). Further, the claim recites steps of generating…at least one prompt…; and generating…a response… which are extra-solution and WURC activities similar to the above analysis (e.g., generating output based on the mental determination). Thus, the claim is ineligible.
Claim Rejections - 35 USC § 103
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.
Claims 1-8, and 11-20 are rejected under AIA 35 U.S.C. 103 as being unpatentable over Ajmera et al. (Pub. No. US 2025/0181899, filed on November 30, 2023; hereinafter Ajmera) in view of Conway et al. (Pub. No. US 2025/0190459, provisionally file don August 10, 2023 based on provisional applications No. 63/518853, 63/518856, and 63/518875; hereinafter Conway).
Regarding claims 1, and 15, Ajmera clearly shows and discloses a computer-implemented method (Abstract); and a computing system, comprising: one or more processors; and one or more computer-readable storage media that store instructions that, when executed by the one or more processors, cause the one or more processors to perform operations (Figure 5), the operations comprising:
storing, by a computing system and for each of a plurality of computing tools, data associated with at least one synthetic query (A row for a particular entity may be converted to markdown language, such that each relationship contained in the database table for the entity is stored in a markdown document, [0019]-[0021], [0027]. It is clear that the markdown documents are used to retrieve relationships between entities within a relational database) generated by one or more machine-learned sequence processing models based at least in part on tool documentation for said each computing tool (The embedding module 146 stores vector representations of the chunks of the markdown documents in the embedding database 130. In some embodiments, the documents and/or data may be associated with an application or suite of applications including an instance running on the client device from which the prompt is entered. Thus, data or metadata generated by or related to various applications belonging to a suite of applications can be chunked and embedded into the embedding database, [0029]);
determining, by the computing system, a subset of the plurality of computing tools that are relevant to a particular user query based at least in part on the data associated with the at least one synthetic query for each of the plurality of computing tools (Such data may be especially relevant to a particular prompt, but unknown to the user or model and/or unavailable using any public or non-recent data source. By generating embeddings based on markup documents and retrieving such embeddings from the embedding database 130, a prompt may be augmented to result in improved generation using a language model due to the particularly relevant data from which the stored embeddings are generated which was not previously available to the model, [0029]. The one or more embeddings for the prompt are passed to the matching service 275, which determines the closest, most similar, or most relevant embeddings from those stored in the persistence service 250, [0041]);
generating, by the computing system, at least one prompt for at least one machine-learned sequence processing model, the at least one prompt including the particular user query (In the example of FIG. 1, the prompt generator 126 receives the embeddings from the embedding selector 124 and generates an augmented prompt based on the input prompt and the stored embeddings that are retrieved by the embedding selector 126, [0030]) and a processing result from each of the subset of the plurality of computing tools in response to the particular user query (The prompt may be augmented by the server device by retrieving embeddings from the embedding database and adding content (e.g., text) corresponding to the retrieved embeddings to the prompt to generate an augmented prompt, [0052]. The workflow 300 may then proceed to stage 360 where a response to the prompt is generated, [0053]); and
generating, by the computing system, a response to the particular user query based at least in part on an output of the at least one machine-learned sequence processing model in response to the at least one prompt (Using augmented prompts, an improved result can be generated using a generative large language model, [0031]. A result of the language model 150 may be received by the result handler 128. The result handler 128 can perform various actions based on the result. For example, the result may be sent to the client device 110, such as being received by the input module 102. The result may then be provided to the output module and output to a client, such as by being displayed on a monitor connected to the client device, [0032]).
Conway then alternatively or additionally discloses:
storing, by a computing system and for each of a plurality of computing tools, data associated with at least one synthetic query generated by one or more machine-learned sequence processing models (At step 252, source data (e.g., a document corpus) is provided to the AI development system 200. At step 254, the KB development facility 213 creates multiple versions of a knowledge base (“candidate KBs”) based on the document corpus. At step 256, the synthetic evaluation data facility 234 generates synthetic user inputs based on the content of the candidate KBs. The synthetic user inputs are generated by prompting a generative model (e.g., an LLM) to identify questions relevant to documents or information in the candidate KBs, [0156]-[0157]) based at least in part on tool documentation for said each computing tool (The corpus of source data may be relevant to a user, an organization, a knowledge domain, one or more topics, etc. The source data may include any suitable type of data (e.g., text, audio, image, video, time-series, etc.), [0115]. It is obvious that the source data can be related to tool documentation);
determining, by the computing system, a subset of the plurality of computing tools that are relevant to a particular user query based at least in part on the data associated with the at least one synthetic query for each of the plurality of computing tools (The completions may include any suitable type of data (e.g., text, audio, image, video, time-series, etc.). This approach is referred to as retrieval-augmented generation because the prompt construction facility 120 (more specifically, the knowledge base search facility 125 of the prompt construction facility 120) retrieves information from the knowledge base 130 and uses that information to augment the user input 110, thereby generating the constructed prompt, [0116]-[0117]);
generating, by the computing system, at least one prompt for at least one machine-learned sequence processing model, the at least one prompt including the particular user query and a processing result from each of the subset of the plurality of computing tools in response to the particular user query (The prompt construction facility 120 may receive user input 110 (e.g., a query or user-generated prompt) to the Gen AI system 100, construct a prompt 135 based on the user input 110 and the knowledge base 130. The generative model 140 generates and outputs generated content 150 (e.g., completions). The completions may include any suitable type of data (e.g., text, audio, image, video, time-series, etc.). This approach is referred to as retrieval-augmented generation because the prompt construction facility 120 (more specifically, the knowledge base search facility 125 of the prompt construction facility 120) retrieves information from the knowledge base 130 and uses that information to augment the user input 110, thereby generating the constructed prompt, [0116]-[0117]).
It would have been obvious to an ordinary person skilled in the art at the time of the invention was effectively filed to incorporate the teachings of Conway with the teachings of Ajmera for the purpose of augmenting a user prompt with context data associated with a processing application and using the augmented prompt and context data to enhance retrieved results responsive to the user prompt.
Regarding claims 2, and 16, Ajmera further discloses:
providing, by the computing system to the one or more machine-learned sequence processing models for each of the plurality of computing tools, at least one synthetic query generation request based at least in part on tool documentation associated with said each computing tool (Using markdown language, the data representing entities and relationships in such a relational database is converted to one or more natural language documents. These natural language documents may be chunked or further processed in some cases. The natural language documents or chunks are used to generate vector embeddings of the natural language of the document, which are then stored in a vector store or vector embedding database, [0020]); and
obtaining, by the computing system from the one or more machine-learned sequence processing models for each computing tool, at least one synthetic query that can be processed by said each computing tool (The markdown module 142 receives data and converts the data to one or more markdown documents, or other markup format language documents. In this context, a markup language may be a template for conveying entities and relationships in a predictable or machine-readable way, [0027]. Data or metadata generated by or related to various applications belonging to a suite of applications can be chunked and embedded into the embedding database, [0029]).
Regarding claims 3, and 17, Ajmera further discloses providing, to the one or more machine-learned sequence processing models for each of the plurality of computing tools, the at least one synthetic query generation request comprises:
providing at least one prompt to the one or more machine-learned sequence processing models, the at least one prompt including the tool documentation associated with said each computing tool and a request to generate at least one synthetic query that can be processed by said each computing tool (The markdown module 142 receives data and converts the data to one or more markdown documents, or other markup format language documents. In this context, a markup language may be a template for conveying entities and relationships in a predictable or machine-readable way, [0027]. Data or metadata generated by or related to various applications belonging to a suite of applications can be chunked and embedded into the embedding database, [0029]).
Regarding claims 4, and 18, Conway then discloses providing, by the computing system to the one or more machine-learned sequence processing models for each of the plurality of computing tools, the at least one synthetic query generation request based at least in part on the tool documentation associated with said each computing tool, comprises:
providing, to the one or more machine-learned sequence processing models for at least one of the plurality of computing tools (The knowledge base 130 contains a representation of information contained in a corpus of source data. The corpus of source data may be relevant to a user, an organization, a knowledge domain, one or more topics, etc. In some embodiments, the KB 130 also includes the source data from which the KB's information representation is derived. The source data may include any suitable type of data (e.g., text, audio, image, video, time-series, etc.), [0115]. At step 252, source data (e.g., a document corpus) is provided to the AI development system 200. At step 254, the KB development facility 213 creates multiple versions of a knowledge base (“candidate KBs”) based on the document corpus, [0156]), a plurality of synthetic query generation requests (At step 256, the synthetic evaluation data facility 234 generates synthetic user inputs based on the content of the candidate KBs. Some non-limiting examples of techniques for generating synthetic evaluation data (e.g., user inputs). The synthetic user inputs are generated by prompting a generative model (e.g., an LLM) to identify questions relevant to documents or information in the candidate KBs, [0156]); and
varying a temperature of the one or more machine-learned sequence processing models for at least one of the plurality of synthetic query generation requests (Hyperparameters are external configuration variables that control or guide machine learning model training. In other words, hyperparameters are parameters that control the learning process and thereby influence the ultimate structure of the model and the learned values of the model parameters. Many hyperparameters are used to guide the training of DNNs, such as the size (number of layers and number of units per layer), the learning rate (e.g., a tuning parameter in an optimization algorithm that determines the step size at each iteration while moving toward a minimum of a loss function), and initial weights of model parameters, [0081]-[0082], [0086]-[0087]).
Regarding claims 5, and 19, Conway further discloses storing data associated with the at least one synthetic query generated by the one or more machine-learned sequence processing models in response to the tool documentation for said each computing tool, comprises: storing the tool documentation for said each computing tool with the at least one synthetic query generated by the one or more machine-learned sequence processing models in response to the tool documentation for said each computing tool (the Gen AI system 100 includes a prompt construction facility 120, a knowledge base (KB) 130, and a generative model 140. The knowledge base 130 contains a representation of information contained in a corpus of source data. The corpus of source data may be relevant to a user, an organization, a knowledge domain, one or more topics, etc. In some embodiments, the KB 130 also includes the source data from which the KB's information representation is derived. The source data may include any suitable type of data (e.g., text, audio, image, video, time-series, etc.). The knowledge base 130 includes a structured representation (e.g., a vector representation, for example, vector embeddings) of information extracted from the source data, [0115]-[0117]).
Regarding claims 6, and 20, Ajmera further discloses:
encoding, into an embedding space, the at least one synthetic query generated by the one or more machine-learned sequence processing models in response to the tool documentation for said each computing tool (An embedding generally refers to a vector representation of an entity that represents the entity as a vector in n-dimensional space such that similar entities are represented by vectors that are close to one another in the n-dimensional space, [0038]);
wherein storing data associated with the at least one synthetic query generated by the one or more machine-learned sequence processing models in response to the tool documentation for said each computing tool comprises, storing at least one embedding of the at least one synthetic query (the embedding module 146 stores vector representations of the chunks of the markdown documents in the embedding database 130. In some embodiments, the documents and/or data may be associated with an application or suite of applications including an instance running on the client device from which the prompt is entered. Thus, data or metadata generated by or related to various applications belonging to a suite of applications can be chunked and embedded into the embedding database, [0029]).
Regarding claim 7, Ajmera further discloses the subset of the plurality computing tools includes less than all of the plurality of computing tools (the embedding module 146 stores vector representations of the chunks of the markdown documents in the embedding database 130. In some embodiments, the documents and/or data may be associated with an application or suite of applications including an instance running on the client device from which the prompt is entered. Thus, data or metadata generated by or related to various applications belonging to a suite of applications can be chunked and embedded into the embedding database, [0029]. The one or more embeddings for the prompt are passed to the matching service 275, which determines the closest, most similar, or most relevant embeddings from those stored in the persistence service 250, [0041]).
Regarding claim 8, Ajmera further discloses the one or more machine-learned sequence processing models includes a first sequence processing model; and the at least one machine-learned sequence processing model includes the first sequence processing model (Embeddings for the chunked documents and/or other data can be generated from the markdown documents by providing the markdown document to the embeddings model 235 via the embeddings API 230 and the ecosystem server 220. For example, embeddings model 235 may be a machine learning model that is trained to generate an n-dimensional vector representation of a set of input data, [0038]).
Regarding claim 11, Ajmera further discloses the one or more machine-learned sequence processing models includes a first large language model (Embeddings model 235 may be, for example, a neural network or other type of machine learning model that learns a representation (embedding) for an entity through a training process that trains the neural network based on a data set, such as a plurality of features of a plurality of entities. In one example, the embedding model comprises a Bidirectional Encoder Representations from Transformer (BERT) model, which involves the use of masked language modeling to determine embeddings, [0038]).
Regarding claim 12, Ajmera further discloses the at least one machine-learned sequence processing model includes a first large language model (Embeddings model 235 may be, for example, a neural network or other type of machine learning model that learns a representation (embedding) for an entity through a training process that trains the neural network based on a data set, such as a plurality of features of a plurality of entities. In one example, the embedding model comprises a Bidirectional Encoder Representations from Transformer (BERT) model, which involves the use of masked language modeling to determine embeddings, [0038]).
Regarding claim 13, Ajmera clearly shows and discloses a computer-implemented method (Abstract), comprising:
providing, by a computing system to one or more machine-learned sequence processing models for each of a plurality of computing tools, at least one synthetic query generation request based at least in part on tool documentation associated with said each computing tool (Using markdown language, the data representing entities and relationships in such a relational database is converted to one or more natural language documents. These natural language documents may be chunked or further processed in some cases. The natural language documents or chunks are used to generate vector embeddings of the natural language of the document, which are then stored in a vector store or vector embedding database, [0019]-[0021]);
obtaining, by the computing system from the one or more machine-learned sequence processing models for each of the plurality of computing tools, at least one synthetic query that can be processed by said each computing tool of the plurality of computing tools (The markdown module 142 receives data and converts the data to one or more markdown documents, or other markup format language documents. In this context, a markup language may be a template for conveying entities and relationships in a predictable or machine-readable way, [0027]. Data or metadata generated by or related to various applications belonging to a suite of applications can be chunked and embedded into the embedding database, [0029]. It is clear that the markdown documents are used to retrieve relationships between entities within a relational database);
storing, by the computing system, data associated with the at least one synthetic query for each of the plurality of computing tools (the embedding module 146 stores vector representations of the chunks of the markdown documents in the embedding database 130. In some embodiments, the documents and/or data may be associated with an application or suite of applications including an instance running on the client device from which the prompt is entered. Thus, data or metadata generated by or related to various applications belonging to a suite of applications can be chunked and embedded into the embedding database, [0029]); and
processing, by the computing system, a particular user query for at least one machine-learned sequence processing model based at least in part on the data associated with the at least one synthetic query for each of the plurality of computing tools (In the example of FIG. 1, the prompt generator 126 receives the embeddings from the embedding selector 124 and generates an augmented prompt based on the input prompt and the stored embeddings that are retrieved by the embedding selector 126, [0030]. The prompt may be augmented by the server device by retrieving embeddings from the embedding database and adding content (e.g., text) corresponding to the retrieved embeddings to the prompt to generate an augmented prompt, [0052]. The workflow 300 may then proceed to stage 360 where a response to the prompt is generated, [0053]).
Conway then alternatively or additionally discloses:
providing, by a computing system to one or more machine-learned sequence processing models for each of a plurality of computing tools, at least one synthetic query generation request (At step 256, the synthetic evaluation data facility 234 generates synthetic user inputs based on the content of the candidate KBs. The synthetic user inputs are generated by prompting a generative model (e.g., an LLM) to identify questions relevant to documents or information in the candidate KBs, [0156]-[0157]) based at least in part on tool documentation associated with said each computing tool (The corpus of source data may be relevant to a user, an organization, a knowledge domain, one or more topics, etc. The source data may include any suitable type of data (e.g., text, audio, image, video, time-series, etc.), [0115]. It is obvious that the source data can be related to tool documentation)
obtaining, by the computing system from the one or more machine-learned sequence processing models for each of the plurality of computing tools, at least one synthetic query that can be processed by said each computing tool of the plurality of computing tools (At step 252, source data (e.g., a document corpus) is provided to the AI development system 200. At step 254, the KB development facility 213 creates multiple versions of a knowledge base (“candidate KBs”) based on the document corpus. At step 256, the synthetic evaluation data facility 234 generates synthetic user inputs based on the content of the candidate KBs. The synthetic user inputs are generated by prompting a generative model (e.g., an LLM) to identify questions relevant to documents or information in the candidate KBs, [0156]-[0157]);
storing, by the computing system, data associated with the at least one synthetic query for each of the plurality of computing tools (the Gen AI system 100 includes a prompt construction facility 120, a knowledge base (KB) 130, and a generative model 140. The knowledge base 130 contains a representation of information contained in a corpus of source data. The corpus of source data may be relevant to a user, an organization, a knowledge domain, one or more topics, etc. In some embodiments, the KB 130 also includes the source data from which the KB's information representation is derived. The source data may include any suitable type of data (e.g., text, audio, image, video, time-series, etc.). The knowledge base 130 includes a structured representation (e.g., a vector representation, for example, vector embeddings) of information extracted from the source data, [0115]-[0117]); and
processing, by the computing system, a particular user query for at least one machine-learned sequence processing model based at least in part on the data associated with the at least one synthetic query for each of the plurality of computing tools (The prompt construction facility 120 may receive user input 110 (e.g., a query or user-generated prompt) to the Gen AI system 100, construct a prompt 135 based on the user input 110 and the knowledge base 130. The generative model 140 generates and outputs generated content 150 (e.g., completions). The completions may include any suitable type of data (e.g., text, audio, image, video, time-series, etc.). This approach is referred to as retrieval-augmented generation because the prompt construction facility 120 (more specifically, the knowledge base search facility 125 of the prompt construction facility 120) retrieves information from the knowledge base 130 and uses that information to augment the user input 110, thereby generating the constructed prompt, [0116]).
It would have been obvious to an ordinary person skilled in the art at the time of the invention was effectively filed to incorporate the teachings of Conway with the teachings of Ajmera for the purpose of augmenting a user prompt with context data associated with a processing application and using the augmented prompt and context data to enhance retrieved results responsive to the user prompt.
Regarding claim 14, Ajmera further discloses processing, by the computing system, the particular user query, comprises:
determining, by the computing system, a subset of the of the plurality of computing tools that are relevant to the particular user query based at least in part on the data associated with the at least one synthetic query for each of the plurality of computing tools (the embedding module 146 stores vector representations of the chunks of the markdown documents in the embedding database 130. In some embodiments, the documents and/or data may be associated with an application or suite of applications including an instance running on the client device from which the prompt is entered. Thus, data or metadata generated by or related to various applications belonging to a suite of applications can be chunked and embedded into the embedding database, [0029]. The one or more embeddings for the prompt are passed to the matching service 275, which determines the closest, most similar, or most relevant embeddings from those stored in the persistence service 250, [0041]);
generating, by the computing system, at least one prompt for the at least one machine-learned sequence processing model, the at least one prompt including the particular user query (In the example of FIG. 1, the prompt generator 126 receives the embeddings from the embedding selector 124 and generates an augmented prompt based on the input prompt and the stored embeddings that are retrieved by the embedding selector 126, [0030]) and a processing result from each of the subset of the plurality of computing tools in response to the particular user query (The prompt may be augmented by the server device by retrieving embeddings from the embedding database and adding content (e.g., text) corresponding to the retrieved embeddings to the prompt to generate an augmented prompt, [0052]. The workflow 300 may then proceed to stage 360 where a response to the prompt is generated, [0053]); and
generating, by the computing system, a response to the particular user query based at least in part on an output of the at least one machine-learned sequence processing model in response to the at least one prompt (Using augmented prompts, an improved result can be generated using a generative large language model, [0031]. A result of the language model 150 may be received by the result handler 128. The result handler 128 can perform various actions based on the result. For example, the result may be sent to the client device 110, such as being received by the input module 102. The result may then be provided to the output module and output to a client, such as by being displayed on a monitor connected to the client device, [0032]).
Conway then alternatively or additionally discloses:
determining, by the computing system, a subset of the plurality of computing tools that are relevant to a particular user query based at least in part on the data associated with the at least one synthetic query for each of the plurality of computing tools (The completions may include any suitable type of data (e.g., text, audio, image, video, time-series, etc.). This approach is referred to as retrieval-augmented generation because the prompt construction facility 120 (more specifically, the knowledge base search facility 125 of the prompt construction facility 120) retrieves information from the knowledge base 130 and uses that information to augment the user input 110, thereby generating the constructed prompt, [0116]-[0117]);
generating, by the computing system, at least one prompt for at least one machine-learned sequence processing model, the at least one prompt including the particular user query and a processing result from each of the subset of the plurality of computing tools in response to the particular user query (The prompt construction facility 120 may receive user input 110 (e.g., a query or user-generated prompt) to the Gen AI system 100, construct a prompt 135 based on the user input 110 and the knowledge base 130. The generative model 140 generates and outputs generated content 150 (e.g., completions). The completions may include any suitable type of data (e.g., text, audio, image, video, time-series, etc.). This approach is referred to as retrieval-augmented generation because the prompt construction facility 120 (more specifically, the knowledge base search facility 125 of the prompt construction facility 120) retrieves information from the knowledge base 130 and uses that information to augment the user input 110, thereby generating the constructed prompt, [0116]-[0117]).
Claims 9-10 are rejected under AIA 35 U.S.C. 103 as being unpatentable over Ajmera in view of Conway and further in view of Huang et al. (Pub. No. US 2025/0111151, filed on September 28, 2023; hereinafter Huang).
Regarding claim 9, Huang then discloses determining, by the computing system the subset of the plurality of computing tools that are relevant to the particular user query, comprises: performing at least one sparse similarity-based retrieval method to compare the particular user query with the data associated with the at least one synthetic query generated by one or more machine-learned sequence processing models in response to tool documentation for said each computing tool (Generative machine learning system 110 may implement a retrieval augmentation pipeline or workflow to perform the natural language request 102. For example, data search 120 may implement sparse retrieval to access data repository index 130, which includes document portions 132 and document metadata 134, [0019]).
It would have been obvious to an ordinary person skilled in the art at the time of the invention was effectively filed to incorporate the teachings of Huang with the teachings of Ajmera, as modified by Conway, for the purpose of using generative machine learning models to perform various natural language processing tasks applicable to a number of different systems, services, or applications to provide or generate data sets for access by natural language generative application services.
Regarding claim 10, Huang further discloses determining, by the computing system the subset of the plurality of computing tools that are relevant to the particular user query, comprises: performing at least one dense similarity-based retrieval method to compare the particular user query with the data associated with the at least one synthetic query generated by one or more machine-learned sequence processing models in response to tool documentation for said each computing tool (Generative machine learning system 110 may implement a retrieval augmentation pipeline or workflow to perform the natural language request 102. For example, data search 120 may implement dense retrieval to access data repository index 130, which includes document portions 132 and document metadata 134, [0019]).
Pertinent Prior Art
The following references are considered relevant to the claims:
Deepak et al. (Pub. No. US 2025/0265253) teaches fetching relevant context responsive to a particular query initially submitted into a large language model without additional context; determining that an output from the large language model for the particular query is an unacceptable output; providing, responsive to the unacceptable output, a select portion of the relevant context to the large language model for a subsequent output; and increasing, progressively and responsive to subsequent unacceptable outputs from the large language model, an amount of context in each subsequent portion iteratively provided to the large language model until an acceptable output is achieved.
Buniatyan (Pat. No. US 12182125) teaches trained embedding mappings for improved retrieval augmented generation wherein a system can maintain a dataset comprising a first set of embeddings corresponding to a first embeddings space and stored in association with a set of query results for the first set of embeddings. The set of query results can correspond to a second embeddings space. The system can train a transformation data structure using the first set of embeddings and the set of query results. The transformation data structure can be used to transform the first set of embeddings to the second embeddings space. The system can execute a search operation for the second embeddings space by applying the transformation data structure to a second set of embeddings corresponding to the first embeddings space.
Contact Information
Any inquiry concerning this communication or earlier communications from the Examiner should be directed to Son Hoang whose telephone number is (571) 270-1752. The Examiner can normally be reached on Monday – Friday (7:00 AM – 4:00 PM).
If attempts to reach the Examiner by telephone are unsuccessful, the Examiner’s supervisor, Sherief Badawi can be reached on (571) 272-9782. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of an application may be obtained from the Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from either Private PAIR or Public PAIR. Status information for unpublished applications is available through Private PAIR only. For more information about the PAIR system, see http://pair-direct.uspto.gov. Should you have questions on access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative or access to the automated information system, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/SON T HOANG/ Primary Examiner, Art Unit 2169 November 25, 2025