Systems and methods for generating customized ai models

Patent metadata

Field	Value
Publication number	US-2025363421-A1
Application number	US-202519288330-A
Country	US
Kind code	A1
Filing date	Aug 1, 2025
Priority date	Sep 25, 2023
Publication date	Nov 27, 2025
Grant date	—

Abstract

While AI models, like large language models, are powerful tools with multiple applications, they can be complex to use and can require a lot of resources to operate. The disclosed systems and methods provide tools to generate customized models (or AI agents) that are configured with features like tailored knowledge, capabilities, and instructions that make them faster, more efficient, and use less computational resources. AI agents may offer several technical advantages of improved efficiency, resource use, and connectivity. This disclosure describes systems and methods to configure, evaluate, generate, and deploy the custom models that can more efficiently run specific tasks. Disclosed systems and methods are configured to, for example, receive a query to generate a custom model, generate the AI agent custom model with the information in the query, and then resolve user queries more efficiently using the custom model.

First claim

What is claimed is:

1 . A system comprising:
at least one memory storing instructions; and at least one processor configured to execute the instructions to perform operations comprising:
receiving, via an interface, a first prompt for building a first model, the prompt comprising model instructions;
processing the prompt with a second model, the second model comprising a language model with an instruction sequence configuring the second model to function as a model builder;
configuring, using the second model, the first model based on the prompt, wherein configuring the first model comprises generating a set of instructions to tune a base model;
generating, using the second model, a first response to the first prompt and a model preview, the model preview comprising a first model name, the first model name being generated by the second model based on the first prompt;
receiving, via the interface, a second prompt in response to the first response and after generating the model preview, the second prompt comprising updating instructions to tune the base model;
processing the second prompt with the second model and updating the first model based on the second prompt, updating the first model comprising modifying the set of instructions for the base model based on the second prompt; and
storing the set of instructions and associating the first model with an account.

2 . The system of claim 1 , wherein the instruction sequence comprises steps of:
determining the first model name based on the first prompt; generating a picture based on the first prompt; and generating the first response, the first response walking the user through refining context.

3 . The system of claim 2 , wherein the instruction sequence comprises the instruction of going through all steps in the instruction sequence in order.

4 . The system of claim 2 , wherein the instruction sequence further comprises an instruction to enter an iterative refinement mode after receiving prompts.

5 . The system of claim 1 , wherein the operations further comprise:
generating, using the second model, a second response to the second prompt; receiving, via the interface, a third prompt in response to the second response, the third prompt comprising updating instructions to tune the base model; and processing the third prompt with the second model and updating the first model based on the third prompt, wherein updating the first model comprises modifying the set of instructions for the base model.

6 . The system of claim 1 , wherein the operations further comprise:
receiving, via the interface, a test prompt after receiving the second prompt; processing, using the first model, the test prompt; and generating, using the first model, a test response to the test prompt.

7 . The system of claim 6 , wherein:
the interface comprises a first prompt input and a second prompt input; the operations further comprise transmitting data to display the model preview in the interface; the first prompt and the second prompt are received from the first prompt input element; the test prompt is received from the second prompt input element.

8 . The system of claim 6 , wherein the operations further comprise transmitting, via the interface, the first response, the second response, and the test response.

9 . The system of claim 1 , wherein:
the first prompt comprises a user defined knowledge base and a user defined capability; configuring the first model based on the prompt comprises establishing connectivity with an API or plugin, and finetuning the base model using the user defined knowledge base.

10 . The system of claim 1 , wherein the operations further comprise generating, using the second model, refining prompts seeking parameters for at least one of: an extraction of data from the knowledge base, a template, an expected output, or an interface configuration.

11 . A computer-implemented method comprising:
receiving, via an interface, a first prompt for building a custom model, the first prompt comprising model instructions; processing the first prompt with a language model configured as a model builder with an instruction sequence; configuring the custom model by generating a set of instructions to tune a base model using the model builder based on the first prompt; generating a first response to the prompt and a model preview using the model builder, the model preview comprising a custom model name based on the first prompt; receiving, via the interface, a second prompt in response to the first response, the second prompt comprising second model instructions for building the custom model; processing the second prompt with the model builder and updating the set of instructions based on the second prompt; and storing the set of instructions and associating the custom model with an account.

12 . The method of claim 11 , further comprising:
receiving a user prompt after associating the custom model with the account; determining the user prompt is associated with the account; and resolving the user prompt using the custom model.

13 . The method of claim 12 , wherein:
configuring the custom model comprises generating a custom interface for interacting with the custom model; and the user prompt is received through the model interface.

14 . The method of claim 11 , wherein the instruction sequence comprises:
determine the custom model name based on the first prompt; and generate a picture based on the first prompt.

15 . The method of claim 11 , wherein the instruction sequence comprises steps of:
generate the first response as a guiding question for refining context; and avoid responses requesting to confirm values for model description.

16 . The method of claim 11 , wherein the instruction sequence comprises an instruction of going through all steps in the instruction sequence in order and without skipping.

17 . The method of claim 11 , wherein:
the instruction sequence comprises an instruction to enter an iterative refinement mode after receiving prompts; and the method further comprises generating, using the model builder, refining prompts seeking parameters for at least one of: an extraction of data from the knowledge base, a template, an expected output, or an interface configuration.

18 . The method of claim 11 , further comprising:
generating, using the model builder, a second response to the second prompt; receiving a third prompt in response to the second response, the third prompt comprising updating instructions to tune the base model; and processing the third prompt with the model builder and modifying the set of instructions.

19 . The method of claim 11 , further comprising:
receiving, via the interface, a test prompt after generating the model preview; processing, using the custom model, the test prompt; and generating, using the custom model, a test response to the test prompt.

20 . A system for configuring a custom AI model, the system comprising:
a network device for hosting an online service; and at least one processor coupled to the network device and configured to:
receive, via an interface, a first prompt for building a custom model, the first prompt comprising model instructions;
process the prompt with a language model configured as a model builder;
configure, using the model builder, the custom model by generating a set of instructions to tune a base model based on the first prompt;
generate, using the model builder, a first response to th

Assignees

• Openai Opco Llc

Inventors

• Turley Nicholas
• Dimson Thomas
• Godement Olivier
• Pokrass Michal

Classifications

• G06N5/02

  Knowledge representation; Symbolic representation · CPC title

• G06F16/3344

  using natural language analysis · CPC title

• G06N5/022

  Knowledge engineering; Knowledge acquisition · CPC title

• G06N20/00 Primary

  Machine learning · CPC title