Granted Patent B1

US 12,596,755 · App. 18/652,804 · Granted Apr 7, 2026

Inventors: Phillip Michael Carter (Redmond, WA); Tatsiana Ramankova (Bonney Lake, WA); Craig Adams Atkinson (Shoreview, MN); Sarah Jeanne Voegeli (Golden, CO)

Assignee: Hound Technology, Inc.

G06F16/9532

Abstract

Claims (89)

1 . A computer-implemented method comprising:

receiving a request to use generative artificial intelligence in association with a query on an observability platform;

generating a first blob of text based on the request;

sending the first blob of text based on the request to a generative artificial intelligence web service;

receiving a response from the generative artificial intelligence web service;

generating a validated response by validating that the response meets a threshold quality level by pre-parsing and processing the response into a test query in a data serialization format and determining the response meets the threshold quality level based on applying one or more query specification rules to the test query, wherein the threshold quality level is determined to be met based, in part, on other quality criteria that is dynamically adjusted based on historical querying data, different database schemas, and user interactions with the observability platform;

executing the validated response as the query on the observability platform; and

responsive to determining that the response does not meet the threshold quality level, modifying the response to comply with one or more query specification rules,

wherein modifying the response to comply with the one or more query specification rules comprises:

pre-parsing and processing the response into a new query in the data serialization format;

identifying one or more portions of the response that prevents validating a parsed data serialization format object into a query format based, in part, on the one or more query specification rules;

modifying the one or more portions of the response to create a modified response, wherein the modifying comprises one or more of: removing the one or more portions of the response, updating the one or more portions, or inserting one or more new portions programmatically; and

executing the modified response as the query in the observability platform.

2 . The computer-implemented method of claim 1 , further comprising:

receiving user feedback on the query on the observability platform.

3 . The computer-implemented method of claim 1 , wherein the first blob of text based on the request is further generated based on:

a description of the one or more query specification rules;

a database schema of the observability platform;

a list of algorithmically-generated relevant names of columns from the database schema;

a list of domain knowledge metadata;

a list of user-provided named examples of queries;

an existing query specification;

a set of examples simulating a conversation with a generative artificial intelligence (AI) web service; and

a set of instructions to produce the query in executable programmatic code.

4 . The computer-implemented method of claim 1 , wherein the threshold quality level is determined to be met based on a set of rules known statically to constitute a valid query, a historical record of executed queries used to generate a likelihood of the response to be validly executed as the query on the observability platform, and wherein the other quality criteria is algorithmically-determined further based on historical usage, dataset schema contents, and at least one of per-user and per-organization behavior on the observability platform.

5 . The computer-implemented method of claim 1 , wherein the executed query on the observability platform is further acted upon, including at least one of naming the executed query, assigning the executed query to a collection of queries, and making the executed query a decision point that causes another action to be performed.

6 . The computer-implemented method of claim 1 , further comprising:

generating a code snippet based, in part, on the one or more query specification rules; and

inserting the code snippet into the modified response.

7 . The computer-implemented method of claim 1 , wherein the data serialization format comprises JavaScript Object Notation (JSON) format and wherein the data serialization format object comprises a JSON object.

8 . A system comprising one or more processors and memory operably coupled with the one or more processors, wherein the memory stores instructions that, in response to execution of the instructions by one or more processors, cause the one or more processors to perform operations including:

receiving a request to use generative artificial intelligence in association with a query on an observability platform;

generating a first blob of text based on the request;

sending the first blob of text based on the request to a generative artificial intelligence web service;

receiving a response from the generative artificial intelligence web service;

generating a validated response by validating that the response meets a threshold quality level by pre-parsing and processing the response into a test query in a data serialization format and determining the response meets the threshold quality level based on applying one or more query specification rules to the test query, wherein the threshold quality level is determined to be met based, in part, on other quality criteria that is dynamically adjusted based on historical querying data, different database schemas, and user interactions with the observability platform;

executing the validated response as the query on the observability platform; and

responsive to determining that the response does not meet the threshold quality level, modifying the response to comply with one or more query specification rules,

wherein modifying the response to comply with the one or more query specification rules comprises:

pre-parsing and processing the response into a new query in the data serialization format;

identifying one or more portions of the response that prevents validating a parsed data serialization format object into a query format based, in part, on the one or more query specification rules;

modifying the one or more portions of the response to create a modified response, wherein the modifying comprises one or more of: removing the one or more portions of the response, updating the one or more portions, or inserting one or more new portions programmatically; and

executing the modified response as the query in the observability platform.

9 . The system of claim 8 , wherein the operations further comprise:

receiving user feedback on the query on the observability platform.

10 . The system of claim 8 , wherein the first blob of text based on the request is further generated based on:

a description of the one or more query specification rules;

a database schema of the observability platform;

a list of algorithmically-generated relevant names of columns from the database schema;

a list of domain knowledge metadata;

a list of user-provided named examples of queries;

an existing query specification;

a set of examples simulating a conversation with a generative artificial intelligence (AI) web service; and

a set of instructions to produce the query in executable programmatic code.

11 . The system of claim 8 , wherein the threshold quality level is determined to be met based on a set of rules known statically to constitute a valid query, a historical record of executed queries used to generate a likelihood of the response to be validly executed as the query on the observability platform, and wherein the other quality criteria is algorithmically-determined further based on historical usage, dataset schema contents, and at least one of per-user and per-organization behavior on the observability platform.

12 . The system of claim 8 , wherein the executed query on the observability platform is further acted upon, including at least one of naming the executed query, assigning the executed query to a collection of queries, and making the executed query a decision point that causes another action to be performed.

13 . The system of claim 8 , wherein the operations further comprise:

generating a code snippet based, in part, on the one or more query specification rules; and

inserting the code snippet into the modified response.

14 . The system of claim 8 , wherein the data serialization format comprises JavaScript Object Notation (JSON) format and wherein the data serialization format object comprises a JSON object.

15 . One or more non-transitory computer-readable storage media storing one or more sequences of instructions which, when executed using one or more processors, cause the one or more processors to execute a method comprising:

receiving a request to use generative artificial intelligence in association with a query on an observability platform;

generating a first blob of text based on the request;

sending the first blob of text based on the request to a generative artificial intelligence web service;

receiving a response from the generative artificial intelligence web service;

generating a validated response by validating that the response meets a threshold quality level by pre-parsing and processing the response into a test query in a data serialization format and determining the response meets the threshold quality level based on applying one or more query specification rules to the test query, wherein the threshold quality level is determined to be met based, in part, on other quality criteria that is dynamically adjusted based on historical querying data, different database schemas, and user interactions with the observability platform;

executing the validated response as the query on the observability platform; and

responsive to determining that the response does not meet the threshold quality level, modifying the response to comply with one or more query specification rules,

wherein modifying the response to comply with the one or more query specification rules comprises:

pre-parsing and processing the response into a new query in the data serialization format;

identifying one or more portions of the response that prevents validating a parsed data serialization format object into a query format based, in part, on the one or more query specification rules;

modifying the one or more portions of the response to create a modified response, wherein the modifying comprises one or more of: removing the one or more portions of the response, updating the one or more portions, or inserting one or more new portions programmatically; and

executing the modified response as the query in the observability platform.

16 . The one or more non-transitory computer-readable storage media of claim 15 , wherein the method further comprises:

receiving user feedback on the query on the observability platform.

17 . The one or more non-transitory computer-readable storage media of claim 15 , wherein the first blob of text based on the request is further generated based on:

a description of the one or more query specification rules;

a database schema of the observability platform;

a list of algorithmically-generated relevant names of columns from the database schema;

a list of domain knowledge metadata;

a list of user-provided named examples of queries;

an existing query specification;

a set of examples simulating a conversation with a generative artificial intelligence (AI) web service; and

a set of instructions to produce the query in executable programmatic code.

18 . The one or more non-transitory computer-readable storage media of claim 15 , wherein the threshold quality level is determined to be met based on a set of rules known statically to constitute a valid query, a historical record of executed queries used to generate a likelihood of the response to be validly executed as the query on the observability platform, and wherein the other quality criteria is algorithmically-determined further based on historical usage, dataset schema contents, and at least one of per-user and per-organization behavior on the observability platform.

19 . The one or more non-transitory computer-readable storage media of claim 15 , wherein the method further comprises:

generating a code snippet based, in part, on the one or more query specification rules; and

inserting the code snippet into the modified response.

20 . The one or more non-transitory computer-readable storage media of claim 15 , wherein the data serialization format comprises JavaScript Object Notation (JSON) format and wherein the data serialization format object comprises a JSON object.

Assignments (1)

Continuity (1)

References Cited (18)