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 .
Response to Amendment
The amendment filed 6/25/2026 has been entered. Claims 1-20 stand amended. Claims 1-20 stand pending.
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.
Claim(s) 1-7, 9-20 is/are rejected under 35 U.S.C. 103 as being unpatentable over Bhattacharya et al. in US Patent № 12,556,559, hereinafter called Bhattacharya, in combination with Poirier et al. in US Patent Application Publication № 2024/0202539, hereinafter called Poirier.
In regard to claim 1, Bhattacharya teaches a system, comprising: a processor (Fig. 1C, element 54) that executes computer executable components stored in memory (Fig. 1C, element 62), wherein the computer executable components comprise:
a tracking component that tracks content changes (i.e. version control, “In a GitOps environment, Git may be viewed as the one and only source of truth. As such, GitOps may require that the desired state of infrastructure (e.g., a customer's cloud deployment) be stored in version control such that the entire audit trail of changes to such infrastructure 60 can be viewed or audited. In a GitOps environment, all changes to infrastructure are embodied as fully traceable commits that are associated with committer information, commit IDs, time stamps, and/or other information.” Column 62 line 55; alternatively or additionally, “In these embodiments, alerts that are generated may be sent to, for example, a chatbot ( e.g., ChatGPT) that can be used to process the alert, including capturing information describing the assets involved, information describing the potential impact of a threat or breach, and so on.” Column 69 line 24), wherein the tracking component monitors and detects changes made to content, including additions, deletions (“For example, if the ingestion and subsequent analysis of audit logs revealed that workloads are deployed and deleted according to some pattern, a customer deploying or deleting a workload in a manner that is inconsistent with the identified pattern may result in an alert being generated or some other remediation workflow being initiated.” Column 93 line 47), or updates (i.e. edits in notebook interface, column 112 line 46);
a natural language processing component that interprets a natural language query (“In addition to those interfaces expressly described already, the systems described herein may leverage natural language interfaces to conduct investigations, access alerts, set policies, or facilitate any of the functionality described herein. Such natural language interfaces can include speech-to-text 30 interfaces, chatbots such as ChatGPT, Natural-language user interfaces (LUI or NLUI), or some other interface that includes natural language processing capabilities.” Column 69 line 25)
an artificial intelligence component that contextualizes, based at least in part on the tracked content changes (“In some embodiments, portions of the data describing activity associated with the anomaly detection framework or the cloud deployment may also be provided as input to the model. For example, such portions of the data may describe a current state of the cloud deployment or a recent history of activity with respect to the cloud deployment, while the model is trained on data describing older, historical activity.” Column 98 line 25; further, “The natural language interface leverages both knowledge of a current or recent state of the cloud deployment as well as historical domain knowledge relevant to security and anomaly detection framework investigations in order to process the received natural language input and generate queries for requested information.” Column 99 line 10), the query and infers sufficiency of a potential result of the contextualized query (i.e. a confidence score, “For example,
while disambiguating a particular term or phrase, multiple interpretations may be determined or identified, with no particular interpretation having a high enough confidence score to be selected.” Column 99 line 51; alternatively or additionally, “In some embodiments, a request for confirmation to use an alternative may be provided 1602 in response to detecting that the natural language input lacks particular keywords, that the resulting query from the natural language input may result in overly broad, overly narrow, or otherwise less-useful data.” Column 101 line 45).
However, Bhattacharya fails to expressly teach wherein the natural language processing component preprocesses the natural language query with identified key information, wherein the artificial intelligence component determines whether there is enough information in results associated with the contextualized query to identify correct content; and a search component that searches metadata of the potential query result and executes the contextualized query.
Popirier teaches wherein the natural language processing component preprocesses the natural language query with identified key information (“In some embodiments, the filter tool module 508-9 can identify implicit filters based on a query or other input, and those identified implicit filters can be used as part of a structed data retrieval process.” Paragraph 0112; alternatively or additionally, “For example, the chunking module 510 may generate enriched embeddings based on the contextual information, data records, and/or data record segments. An enriched embedding may comprise a vector value based on an embedding vector and the contextual information.” Paragraph 0119)
wherein the artificial intelligence component determines whether there is enough information in results associated with the contextualized query to identify correct content (“The comprehension module 516 can function to process inputs to determine results (e.g., "answers"), determine rationales for results, and determine whether the comprehension module 516 needs more information to determine results. The comprehension module 516 may output information ( e.g., results or additional queries) in a natural language format or machine language format. In some implementations, features of one or more models of the comprehension module define conditions or functions that determine if more information is needed to satisfy the initial input or if there is enough information to satisfy the initial input.” Paragraph 0122);
and a search component that searches metadata of the potential query result and executes the contextualized query (“For example, one or more of the agent modules 506 may obtain/output a document (or segment(s) thereof) or related information (e.g., text summary or translation), another agent module 506 may obtain/output a database table, and the like. The orchestrator 504 may then use one or more machine learning models ( e.g., a large language model and/or another machine learning model) to combine the outputs/results into a unified output ( e.g., having a common data format, such as natural language).” Paragraph 0083).
It would have been obvious before the effective filing date of the instant invention to modify the change-tracking AI query system taught by Bhattacharya to include the query enrichment and sufficiency determination taught by Poirier. It would have been obvious because it represents the application of a known technique (i.e. the query enrichment taught by Poirier in at least paragraph 0112, and the determination of response sufficiency taught by Poirier in at least paragraph 0122) to a known system (i.e. the change-traching query system taught by Bhattacharya in at least column 98 line 25) ready for improvement to yield only predictable results (i.e. the system can retrieve additional information if there is not enough to generate an answer, as taught by Poirier in at least paragraph 0151)
In regard to claim 2, Bhattacharya further teaches a search component that executes the contextualized query to return matching records based on included data and metadata (“The example method depicted in FIG. 13 also includes providing 1306, based on a response to the query, a response to the natural language input. The response to the query may be generated by providing the query to an API, database software, or other component that selects and/or processes data based on the query and provides the selected and/or processed data as a response to the query.” Column 98 line 50).
In regard to claim 3, Bhattacharya further teaches that the artificial intelligence component determines sufficiency of the query as a function of a confidence score of utility to a user of the query (“For example, while disambiguating a particular term or phrase, multiple interpretations may be determined or identified, with no particular interpretation having a high enough confidence score to be selected.” Column 99 line 51; alternatively or additionally, “In some embodiments, a request for confirmation to use an alternative may be provided 1602 in response to detecting that the natural language input lacks particular keywords, that the resulting query from the natural language input may result in overly broad, overly narrow, or otherwise less-useful data.” Column 101 line 45).
However, Bhattacharya fails to expressly teach including determining whether there is enough information to identify correct content.
Poirier teaches including determining whether there is enough information to identify correct content ((“The comprehension module 516 can function to process inputs to determine results (e.g., "answers"), determine rationales for results, and determine whether the comprehension module 516 needs more information to determine results. The comprehension module 516 may output information ( e.g., results or additional queries) in a natural language format or machine language format. In some implementations, features of one or more models of the comprehension module define conditions or functions that determine if more information is needed to satisfy the initial input or if there is enough information to satisfy the initial input.” Paragraph 0122)
It would have been obvious before the effective filing date of the instant invention to modify the change-tracking AI query system taught by Bhattacharya to include the query enrichment and sufficiency determination taught by Poirier. It would have been obvious because it represents the application of a known technique (i.e. the query enrichment taught by Poirier in at least paragraph 0112, and the determination of response sufficiency taught by Poirier in at least paragraph 0122) to a known system (i.e. the change-tracking query system taught by Bhattacharya in at least column 98 line 25) ready for improvement to yield only predictable results (i.e. the system can retrieve additional information if there is not enough to generate an answer, as taught by Poirier in at least paragraph 0151)
In regard to claim 4, Bhattacharya further teaches that the artificial intelligence component uses situational analysis to determine sufficiency of the query based on utility of the query. (“In some embodiments, a request for confirmation to use an alternative may be provided 1602 in response to detecting that the natural language input lacks particular keywords, that the resulting query from the natural language input may result in overly broad, overly narrow, or otherwise less useful data.” Column 101 line 45).
In regard to claim 5, Bhattacharya further teaches that, upon determining that the query will not result in utility, the artificial intelligence component requests more information to identify the correct content (“In some embodiments, providing 1402 the request for clarification may be performed in response to determining that a particular term or phrase in the natural language input cannot be disambiguated without clarification. For example, while disambiguating a particular term or phrase, multiple interpretations may be determined or identified, with no particular interpretation having a high enough confidence score to be selected. Accordingly, a clarification may be required to select a particular interpretation.” Column 99 line 48).
In regard to claim 6, Bhattacharya further teaches that the artificial intelligence component provides customized recommendations to modify the natural language query and receives a modified query (“In some embodiments, the request for clarification may include a request to select one of multiple interpretations. Continuing with the example above for a natural language input of "Show all active virtual machines," a request for clarification may be provided as "Did you mean 'Show all active virtual machines for my organization' or 'Show all active virtual machines for my group'?" column 100 line 19, wherein a new query may be generated, column 200 line 29).
In regard to claim 7, Bhattacharya further teaches that the customized recommendations are based at least in part upon the tracked content changes, including code repository changes (“For example, assume that an anomaly has been detected in the cloud framework and that, during past instances of this anomaly, one or more users have provided as a natural language input "Show all active virtual machines for my group." Further assume that a natural language input is received that states "Show all active virtual machines." A request for clarification may be provided 1402 as "Did you mean 'Show all active virtual machines for my group?"' Depending on a response to the 15 request (e.g., a "yes" or "no" input), the query may be generated to show all active virtual machines for the user group, or all active virtual machines for the organization as a whole.” Column 100 line 6, wherein a repository that includes code is tracked for changes, “As such, the embodiments described herein may be configured to evaluate various entities for signatures that are indicative of AI generated code, such as the inclusion of libraries typically used by AI code generating tools, programming styles that are common in code that is generated by AI code generating tools, or any other marker that some piece of software was generated by AI code generating tools. In such a way, software that is generated by AI code generating tools (which may be used for the rapid development of malicious code) may be identified and subjected to a higher level of scrutiny as code that is generated in more traditional ways.” Column 69 line 10).
In regard to claim 9, Bhattacharya further teaches that the artificial intelligence component uses named entity recognition to contextualize the query by identifying and categorizing entities mentioned in text. (“In some examples, the queries provided by data processing resources 20 may be configured to direct data store 30 to perform one or more data analytics operations with respect to the data stored within data store 30. These data analytics operations may be with respect to data specific to a particular entity ( e.g., data residing in one or more silos within data store 30 that are associated with a particular customer) and/or data associated with multiple entities. For example, data processing resources 20 may be configured to analyze data associated with a first entity and use the results of the analysis to perform one or more operations with respect to a second entity.” Column 6 line 22, wherein the entities are taught to have names in at least column 9 line 16, and the extraction from text is taught in at least column 102 line 56).
In regard to claim 10, Bhattacharya further teaches that the artificial intelligence component analyzes metadata of the potential query result wherein the metadata comprises comments or commit messages (“In some embodiments, various attributes or metadata associated with alerts may also be used to train the model, including in association with the interactions described above. For example, text descriptions or summaries of particular events or alerts may be used to train or establish associations between particular keywords or descriptors and particular interactions. Such information may then be used to generate queries for alerts or events that may be similar (e.g., sharing similar keywords or descriptions), but not identical, to previously generated alerts.” Column 102 line 56).
In regard to claim 11, Bhattacharya further teaches that the artificial intelligence component uses keyword extraction to contextualize the query wherein the keyword extraction identifies main topics (i.e. the key subject) associated with the query. (“In some embodiments, various attributes or metadata associated with alerts may also be used to train the model, including in association with the interactions described above. For example, text descriptions or summaries of particular events or alerts may be used to train or establish associations between particular keywords or descriptors and particular interactions. Such information may then be used to generate queries for alerts or events that may be similar ( e.g., sharing similar keywords or descriptions), but not identical, to previously generated alerts.” Column 102 line 56).
In regard to claim 12, Bhattacharya teaches a computer-implemented method that utilizes a processor (Fig. 1C, element 54) that executes computer executable components stored in memory (Fig. 1C, element 62) to perform the following acts:
tracking content changes (i.e. version control, “In a GitOps environment, Git may be viewed as the one and only source of truth. As such, GitOps may require that the desired state of infrastructure (e.g., a customer's cloud deployment) be stored in version control such that the entire audit trail of changes to such infrastructure 60 can be viewed or audited. In a GitOps environment, all changes to infrastructure are embodied as fully traceable commits that are associated with committer information, commit IDs, time stamps, and/or other information.” Column 62 line 55; alternatively or additionally, “In these embodiments, alerts that are generated may be sent to, for example, a chatbot ( e.g., ChatGPT) that can be used to process the alert, including capturing information describing the assets involved, information describing the potential impact of a threat or breach, and so on.” Column 69 line 24) including monitoring and detecting additions, deletions (“For example, if the ingestion and subsequent analysis of audit logs revealed that workloads are deployed and deleted according to some pattern, a customer deploying or deleting a workload in a manner that is inconsistent with the identified pattern may result in an alert being generated or some other remediation workflow being initiated.” Column 93 line 47),or updates (i.e. edits in notebook interface, column 112 line 46) associated with content;
interpreting a natural language query (“In addition to those interfaces expressly described already, the systems described herein may leverage natural language interfaces to conduct investigations, access alerts, set policies, or facilitate any of the functionality described herein. Such natural language interfaces can include speech-to-text 30 interfaces, chatbots such as ChatGPT, Natural-language user interfaces (LUI or NLUI), or some other interface that includes natural language processing capabilities.” Column 69 line 25);
contextualizing, based at least in part on the tracked content changes, the query (“In some embodiments, portions of the data describing activity associated with the anomaly detection framework or the cloud deployment may also be provided as input to the model. For example, such portions of the data may describe a current state of the cloud deployment or a recent history of activity with respect to the cloud deployment, while the model is trained on data describing older, historical activity.” Column 98 line 25; further, “The natural language interface leverages both knowledge of a current or recent state of the cloud deployment as well as historical domain knowledge relevant to security and anomaly detection framework investigations in order to process the received natural language input and generate queries for requested information.” Column 99 line 10);
and inferring a potential result of the contextualize query (i.e. an inference of an ambiguous result “For example, while disambiguating a particular term or phrase, multiple interpretations may be determined or identified, with no particular interpretation having a high enough confidence score to be selected.” Column 99 line 51; alternatively or additionally, “In some embodiments, a request for confirmation to use an alternative may be provided 1602 in response to detecting that the natural language input lacks particular keywords, that the resulting query from the natural language input may result in overly broad, overly narrow, or otherwise less-useful data.” Column 101 line 45)).
However, Bhattacharya fails to expressly teach wherein the natural language processing component preprocesses the natural language query with identified key information, wherein the artificial intelligence component determines whether there is enough information in results associated with the contextualized query to identify correct content; and a search component that searches metadata of the potential query result and executes the contextualized query.
Popirier teaches wherein the natural language processing component preprocesses the natural language query with identified key information (“In some embodiments, the filter tool module 508-9 can identify implicit filters based on a query or other input, and those identified implicit filters can be used as part of a structed data retrieval process.” Paragraph 0112; alternatively or additionally, “For example, the chunking module 510 may generate enriched embeddings based on the contextual information, data records, and/or data record segments. An enriched embedding may comprise a vector value based on an embedding vector and the contextual information.” Paragraph 0119)
wherein the artificial intelligence component determines whether there is enough information in results associated with the contextualized query to identify correct content (“The comprehension module 516 can function to process inputs to determine results (e.g., "answers"), determine rationales for results, and determine whether the comprehension module 516 needs more information to determine results. The comprehension module 516 may output information ( e.g., results or additional queries) in a natural language format or machine language format. In some implementations, features of one or more models of the comprehension module define conditions or functions that determine if more information is needed to satisfy the initial input or if there is enough information to satisfy the initial input.” Paragraph 0122);
and a search component that searches metadata of the potential query result and executes the contextualized query (“For example, one or more of the agent modules 506 may obtain/output a document (or segment(s) thereof) or related information (e.g., text summary or translation), another agent module 506 may obtain/output a database table, and the like. The orchestrator 504 may then use one or more machine learning models ( e.g., a large language model and/or another machine learning model) to combine the outputs/results into a unified output ( e.g., having a common data format, such as natural language).” Paragraph 0083).
It would have been obvious before the effective filing date of the instant invention to modify the change-tracking AI query system taught by Bhattacharya to include the query enrichment and sufficiency determination taught by Poirier. It would have been obvious because it represents the application of a known technique (i.e. the query enrichment taught by Poirier in at least paragraph 0112, and the determination of response sufficiency taught by Poirier in at least paragraph 0122) to a known system (i.e. the change-traching query system taught by Bhattacharya in at least column 98 line 25) ready for improvement to yield only predictable results (i.e. the system can retrieve additional information if there is not enough to generate an answer, as taught by Poirier in at least paragraph 0151)
In regard to claim 13, Bhattacharya further teaches executing the contextualized query further comprises returning matching records based on included data and metadata (“The example method depicted in FIG. 13 also includes providing 1306, based on a response to the query, a response to the natural language input. The response to the query may be generated by providing the query to an API, database software, or other component that selects and/or processes data based on the query and provides the selected and/or processed data as a response to the query.” Column 98 line 50).
In regard to claim 14, Bhattacharya further teaches determining sufficiency of the query as a function of a confidence score of utility to a user of the query (“For example, while disambiguating a particular term or phrase, multiple interpretations may be determined or identified, with no particular interpretation having a high enough confidence score to be selected.” Column 99 line 51; alternatively or additionally, “In some embodiments, a request for confirmation to use an alternative may be provided 1602 in response to detecting that the natural language input lacks particular keywords, that the resulting query from the natural language input may result in overly broad, overly narrow, or otherwise less-useful data.” Column 101 line 45).
In regard to claim 15, Bhattacharya further teaches using situational analysis to determine sufficiency of the query (i.e. awareness of the current status, “In some embodiments, portions of the data describing activity associated with the anomaly detection framework or the cloud deployment may also be provided as input to the model. For example, such portions of the data may describe a current state of the cloud deployment or a recent history of activity with respect to the cloud deployment, while the model is trained on data describing older, historical activity.” Column 98 line 25).
However, Bhattacharya fails to expressly teach including determining whether there is enough information to identify correct content.
Poirier teaches including determining whether there is enough information to identify correct content ((“The comprehension module 516 can function to process inputs to determine results (e.g., "answers"), determine rationales for results, and determine whether the comprehension module 516 needs more information to determine results. The comprehension module 516 may output information ( e.g., results or additional queries) in a natural language format or machine language format. In some implementations, features of one or more models of the comprehension module define conditions or functions that determine if more information is needed to satisfy the initial input or if there is enough information to satisfy the initial input.” Paragraph 0122)
It would have been obvious before the effective filing date of the instant invention to modify the change-tracking AI query system taught by Bhattacharya to include the query enrichment and sufficiency determination taught by Poirier. It would have been obvious because it represents the application of a known technique (i.e. the query enrichment taught by Poirier in at least paragraph 0112, and the determination of response sufficiency taught by Poirier in at least paragraph 0122) to a known system (i.e. the change-tracking query system taught by Bhattacharya in at least column 98 line 25) ready for improvement to yield only predictable results (i.e. the system can retrieve additional information if there is not enough to generate an answer, as taught by Poirier in at least paragraph 0151)
In regard to claim 16, Bhattacharya further teaches determining that the query does not contain sufficient information, and issuing a request for more information to identify the correct content. (“In some embodiments, providing 1402 the request for clarification may be performed in response to determining that a particular term or phrase in the natural language input cannot be disambiguated without clarification. For example, while disambiguating a particular term or phrase, multiple interpretations may be determined or identified, with no particular interpretation having a high enough confidence score to be selected. Accordingly, a clarification may be required to select a particular interpretation.” Column 99 line 48).
In regard to claim 17, Bhattacharya further teaches searching metadata of the potential query result further comprises comments or commit messages. (“In some embodiments, various attributes or metadata associated with alerts may also be used to train the model, including in association with the interactions described above. For example, text descriptions or summaries of particular events or alerts may be used to train or establish associations between particular keywords or descriptors and particular interactions. Such information may then be used to generate queries for alerts or events that may be similar (e.g., sharing similar keywords or descriptions), but not identical, to previously generated alerts.” Column 102 line 56).
In regard to claim 18, Bhattacharya further teaches that the customized recommendations are based at least in part upon the tracked content changes, including code repository changes. (“For example, assume that an anomaly has been detected in the cloud framework and that, during past instances of this anomaly, one or more users have provided as a natural language input "Show all active virtual machines for my group." Further assume that a natural language input is received that states "Show all active virtual machines." A request for clarification may be provided 1402 as "Did you mean 'Show all active virtual machines for my group?"' Depending on a response to the 15 request (e.g., a "yes" or "no" input), the query may be generated to show all active virtual machines for the user group, or all active virtual machines for the organization as a whole.” Column 100 line 6, wherein a repository that includes code is tracked for changes, “As such, the embodiments described herein may be configured to evaluate various entities for signatures that are indicative of AI generated code, such as the inclusion of libraries typically used by AI code generating tools, programming styles that are common in code that is generated by AI code generating tools, or any other marker that some piece of software was generated by AI code generating tools. In such a way, software that is generated by AI code generating tools (which may be used for the rapid development of malicious code) may be identified and subjected to a higher level of scrutiny as code that is generated in more traditional ways.” Column 69 line 10).
In regard to claim 19, Bhattacharya further teaches using keyword extraction to contextualize the query wherein the keyword extraction identifies main topics associated with the query. (“In some embodiments, various attributes or metadata associated with alerts may also be used to train the model, including in association with the interactions described above. For example, text descriptions or summaries of particular events or alerts may be used to train or establish associations between particular keywords or descriptors and particular interactions. Such information may then be used to generate queries for alerts or events that may be similar ( e.g., sharing similar keywords or descriptions), but not identical, to previously generated alerts.” Column 102 line 56).
In regard to claim 20, Bhattacharya further teaches a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a processor to cause the processor to:
track content changes (i.e. version control, “In a GitOps environment, Git may be viewed as the one and only source of truth. As such, GitOps may require that the desired state of infrastructure (e.g., a customer's cloud deployment) be stored in version control such that the entire audit trail of changes to such infrastructure 60 can be viewed or audited. In a GitOps environment, all changes to infrastructure are embodied as fully traceable commits that are associated with committer information, commit IDs, time stamps, and/or other information.” Column 62 line 55; alternatively or additionally, “In these embodiments, alerts that are generated may be sent to, for example, a chatbot ( e.g., ChatGPT) that can be used to process the alert, including capturing information describing the assets involved, information describing the potential impact of a threat or breach, and so on.” Column 69 line 24) including additions, deletions (“For example, if the ingestion and subsequent analysis of audit logs revealed that workloads are deployed and deleted according to some pattern, a customer deploying or deleting a workload in a manner that is inconsistent with the identified pattern may result in an alert being generated or some other remediation workflow being initiated.” Column 93 line 47), or updates (i.e. edits in notebook interface, column 112 line 46);
receive a natural language query (“In addition to those interfaces expressly described already, the systems described herein may leverage natural language interfaces to conduct investigations, access alerts, set policies, or facilitate any of the functionality described herein. Such natural language interfaces can include speech-to-text 30 interfaces, chatbots such as ChatGPT, Natural-language user interfaces (LUI or NLUI), or some other interface that includes natural language processing capabilities.” Column 69 line 25);
interpret the query (“In addition to those interfaces expressly described already, the systems described herein may leverage natural language interfaces to conduct investigations, access alerts, set policies, or facilitate any of the functionality described herein. Such natural language interfaces can include speech-to-text 30 interfaces, chatbots such as ChatGPT, Natural-language user interfaces (LUI or NLUI), or some other interface that includes natural language processing capabilities.” Column 69 line 25);
contextualize, based at least in part on the tracked content changes, the query (“In some embodiments, portions of the data describing activity associated with the anomaly detection framework or the cloud deployment may also be provided as input to the model. For example, such portions of the data may describe a current state of the cloud deployment or a recent history of activity with respect to the cloud deployment, while the model is trained on data describing older, historical activity.” Column 98 line 25; further, “The natural language interface leverages both knowledge of a current or recent state of the cloud deployment as well as historical domain knowledge relevant to security and anomaly detection framework investigations in order to process the received natural language input and generate queries for requested information.” Column 99 line 10);
and infer a potential result of the contextualized query (i.e. an inference of an ambiguous result “For example, while disambiguating a particular term or phrase, multiple interpretations may be determined or identified, with no particular interpretation having a high enough confidence score to be selected.” Column 99 line 51; alternatively or additionally, “In some embodiments, a request for confirmation to use an alternative may be provided 1602 in response to detecting that the natural language input lacks particular keywords, that the resulting query from the natural language input may result in overly broad, overly narrow, or otherwise less-useful data.” Column 101 line 45).
However, Bhattacharya fails to expressly teach wherein the natural language processing component preprocesses the natural language query with identified key information, wherein the artificial intelligence component determines whether there is enough information in results associated with the contextualized query to identify correct content; and a search component that searches metadata of the potential query result and executes the contextualized query.
Popirier teaches wherein the natural language processing component preprocesses the natural language query with identified key information (“In some embodiments, the filter tool module 508-9 can identify implicit filters based on a query or other input, and those identified implicit filters can be used as part of a structed data retrieval process.” Paragraph 0112; alternatively or additionally, “For example, the chunking module 510 may generate enriched embeddings based on the contextual information, data records, and/or data record segments. An enriched embedding may comprise a vector value based on an embedding vector and the contextual information.” Paragraph 0119)
wherein the artificial intelligence component determines whether there is enough information in results associated with the contextualized query to identify correct content (“The comprehension module 516 can function to process inputs to determine results (e.g., "answers"), determine rationales for results, and determine whether the comprehension module 516 needs more information to determine results. The comprehension module 516 may output information ( e.g., results or additional queries) in a natural language format or machine language format. In some implementations, features of one or more models of the comprehension module define conditions or functions that determine if more information is needed to satisfy the initial input or if there is enough information to satisfy the initial input.” Paragraph 0122);
and a search component that searches metadata of the potential query result and executes the contextualized query (“For example, one or more of the agent modules 506 may obtain/output a document (or segment(s) thereof) or related information (e.g., text summary or translation), another agent module 506 may obtain/output a database table, and the like. The orchestrator 504 may then use one or more machine learning models ( e.g., a large language model and/or another machine learning model) to combine the outputs/results into a unified output ( e.g., having a common data format, such as natural language).” Paragraph 0083).
It would have been obvious before the effective filing date of the instant invention to modify the change-tracking AI query system taught by Bhattacharya to include the query enrichment and sufficiency determination taught by Poirier. It would have been obvious because it represents the application of a known technique (i.e. the query enrichment taught by Poirier in at least paragraph 0112, and the determination of response sufficiency taught by Poirier in at least paragraph 0122) to a known system (i.e. the change-traching query system taught by Bhattacharya in at least column 98 line 25) ready for improvement to yield only predictable results (i.e. the system can retrieve additional information if there is not enough to generate an answer, as taught by Poirier in at least paragraph 0151)
Claim(s) 8 is/are rejected under 35 U.S.C. 103 as being unpatentable over Bhattacharya and Poirier as applied to claim 1 above, and further in view of Chakraborty et al. in US Patent Application Publication № 12,306,828, hereinafter called Chakraborty.
In regard to claim 8, Bhattacharya and Poirier teach the system of claim 1, as above. However, they fail to expressly teach that the natural language processing component utilizes word embedding to interpret the query, wherein the query is converted into numerical format using the word embedding.
Chakraborty teaches that the natural language processing component utilizes word embedding to interpret the query wherein the query is converted into numerical format using the word embedding. (“Matching may be performed by converting the natural language question received to embeddings. Once converted, those embeddings may be compared to existing embeddings for user defined questions and/or other pre-generated questions, such as similar questions and interesting questions previously generated using an LLM or other generative AI.” Column 4 line 10, wherein mathematical, i.e. numerical, operations are taught to be performed on the embeddings in at least column 4 line 28).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the instant invention to modify the natural-language query processing system which compares current queries with previous queries, as taught by Bhattacharya, to include the conversion of a natural-language query to embeddings to compare to previously-stored queries, as taught by Chakraborty. It would have been obvious because it represents the use of a known technique (i.e. the conversion of a natural language query to an embedding and then comparing using vector comparison, as taught by Chakraborty in column 4, lines 10-37) to improve similar systems (i.e. the natural-language query translation system, as taught by Chakraborty, “The UQR may solve the problem of converting a natural language question to a structured query to be run against a structured database by sending relevant context of the question enriched with relevant metadata to an LLM for query generation”, column 3 line 49, and the query natural-language query translation system taught by Bhattacharya, “The method of FIG. 13 also includes generating 1304 a query corresponding to the natural language input by disambiguating at least a portion of the natural language input based on data describing activity associated with the cloud deployment. The query may be embodied, for example, as a database query expressed in languages such as SQL, an API call, or other query that retrieves information necessary to respond to the natural language input.” Column 95, line 20; interpretation of similar textual information to historical examples is additionally taught in at least column 102 line 56) in the same way (i.e. the use of embeddings to do vector comparison), One would have been motivated to do so in order to provide the most relevant context for the natural language query to the LLM, as taught by Chakraborty in at least column 3 line 62.
Response to Arguments
Applicant’s arguments, see pages 6-10, filed 6/25/2026, with respect to the rejection of claims 1-7 and 9-20 under 35 U.S.C. 102(a)(2), and claim 8 under 35 U.S.C. 103, have been fully considered and are persuasive. Therefore, the rejection has been withdrawn. However, upon further consideration, a new ground(s) of rejection is made in view of Bhattacharya and Poirier. For more information, please refer to the relevant seconds above.
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Lauren Z Ganger whose telephone number is (571)272-0270. The examiner can normally be reached 10:00 AM - 7:30 PM.
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, Ajay Bhatia can be reached at (571) 272-3906. 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.
/VAISHALI SHAH/Primary Examiner, Art Unit 2156
Prosecution Insights