Filtering materials based on user intent capture using large language models

Patent metadata

Field	Value
Publication number	US-2026003858-A1
Application number	US-202519301614-A
Country	US
Kind code	A1
Filing date	Aug 15, 2025
Priority date	Jun 28, 2024
Publication date	Jan 1, 2026
Grant date	—

Abstract

Various embodiments are directed towards techniques for determining materials for computer-generated designs that include generating a query prompt based on an assembly context, transmitting the query prompt to a plurality of large language model (LLM) agents for processing, receiving a plurality of material attribute filters from the plurality of LLM agents, where each LLM generates a different material attribute filter when processing the query prompt, combining the material attribute filters included in the plurality of material attribute filters to produce a material query, querying a material database using the material query to identify at least one potential material to use for a design, evaluating simulation results to determine whether the at least one material is an appropriate material to use for the design.

First claim

What is claimed is:

1 . A computer-implemented method for identifying materials for designs, the method comprising:
receiving design information for a design; generating, via one or more large language models, material-related outputs based on the design information; generating a query based on a combination of the material-related outputs; querying a data source of materials using the query to obtain candidate materials; performing at least one simulation involving the design and one or more of the candidate materials to generate simulation results; and generating an output that identifies one or more materials based on the simulation results.

2 . The computer-implemented method of claim 1 , wherein the one or more large language models comprise a plurality of large language model agents, and each large language model agent included in the plurality of large language model agents corresponds to a different material attribute is configured to generate a material attribute filter in a structured format.

3 . The computer-implemented method of claim 1 , wherein the design information comprises both structured data obtained from a computer-aided design environment and a user-provided description of an intended use of the design.

4 . The computer-implemented method of claim 1 , wherein the simulation results are generated by simulating the design with each candidate material to determine compliance with a plurality of design criteria specified by a user.

5 . The computer-implemented method of claim 1 , wherein generating the output comprises designating at least one candidate material as suitable for the design when the simulation results satisfy specified performance thresholds.

6 . The computer-implemented method of claim 1 , further comprising, when no candidate material satisfies a performance threshold, generating an updated query based on the simulation results and providing the updated query to the one or more large language models.

7 . The computer-implemented method of claim 1 , wherein the data source of materials comprises at least one of a local material library or a remote material database.

8 . The computer-implemented method of claim 1 , wherein the simulation results are generated using a simulation engine external to a computer-aided design environment.

9 . The computer-implemented method of claim 1 , wherein the output comprises a visualization comparing performance metrics for the one or more materials.

10 . The computer-implemented method of claim 1 , wherein the one or more large language models incorporate prior simulation results into subsequent material-related outputs.

11 . One or more non-transitory computer-readable media storing instructions that, when executed by one or more processors, cause the one or more processors to identify materials for designs, by performing the operations of:
receiving design information for a design; generating, via one or more large language models, material-related outputs based on the design information; obtaining candidate materials based on the material-related outputs; performing at least one simulation involving the design and one or more of the candidate materials to generate simulation results; and generating an output that identifies one or more materials based on the simulation results.

12 . The one or more non-transitory computer-readable media of claim 11 , wherein the one or more large language models comprise a plurality of large language model agents, and each large language model agent included in the plurality of large language model agents corresponds to a different material attribute is configured to generate a material attribute filter in a structured format.

13 . The one or more non-transitory computer-readable media of claim 11 , wherein the design information comprises both structured data obtained from a computer-aided design environment and a user-provided description of an intended use of the design.

14 . The one or more non-transitory computer-readable media of claim 11 , wherein the simulation results are generated by simulating the design with each candidate material to determine compliance with a plurality of design criteria specified by a user.

15 . The one or more non-transitory computer-readable media of claim 11 , wherein generating the output comprises designating at least one candidate material as suitable for the design when the simulation results satisfy specified performance thresholds.

16 . The one or more non-transitory computer-readable media of claim 11 , wherein the candidate materials are obtained from a data source of materials comprising multiple material databases accessible via different interfaces.

17 . The one or more non-transitory computer-readable media of claim 11 , wherein the simulation results are generated using a simulation engine located on a different computing system.

18 . The one or more non-transitory computer-readable media of claim 11 , wherein the output includes a visualization of simulation results for the one or more materials.

19 . The one or more non-transitory computer-readable media of claim 11 , wherein the one or more large language models incorporate prior simulation results into subsequent material-related outputs.

20 . A system, comprising:
one or more processors; and one or more memories storing instructions that, when executed by the one or more processors, cause the one or more processors to identify materials for designs, by performing the operations of:
receiving design information for a design;
generating, via one or more large language models, material-related outputs based on the design information;
generating a query based on a combination of the material-related outputs;
querying a data source of materials using the query to obtain candidate materials;
performing at least one simulation involving the design and one or more of the candidate materials to generate simulation results; and
generating an output that identifies one or more materials based on the simulation results.

Assignees

• Autodesk Inc

Inventors

• Harris Andrew John
• Grandi Daniele
• Wannamaker Kendra Ann
• Chen Michael
• Groom Allin Irving

Classifications

• G06F30/20

  Design optimisation, verification or simulation (optimisation, verification or simulation of circuit designs G06F30/30) · CPC title

• G06F16/248

  Presentation of query results · CPC title

• G06F2119/18

  Manufacturability analysis or optimisation for manufacturability · CPC title

• G06F30/27

  using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model · CPC title

• G06F16/2425 Primary

  Iterative querying; Query formulation based on the results of a preceding query · CPC title