Nvidia patents AI system to find software bugs using logs
Nvidia describes a system that uses two stages of AI models to analyze test logs and product logs. The first model extracts context from test failures, which helps the second model pinpoint the root cause in the product logs. The system then generates reports and suggests code fixes.
In the filing’s own words
In various examples, systems and methods for root cause failure identification using multi-stage retrieval-augmented generation are provided. A multi-stage retrieval-augmented generation system may include first model(s) that may extract a first context from test log(s) including a test failure based on first prompt(s) and information retrieved from first database(s). The multi-stage retrieval-augmented generation system may also include second model(s) that may extract a second context from product log(s) corresponding to the test log(s) based on the first context extracted from the test log(s), second prompt(s), and information retrieved from second database(s). Operation(s) for addressing the test failure may be performed based on the second context extracted from the product log(s) includes generating a report that identifies a root cause of at least one test failure, generating proposed change(s) to a test script for the automated testing, and/or generating proposed change(s) to source code for the product under test.
What exactly they claim (claim 1)
1. One or more processors comprising processing circuitry to: generate, using one or more first models, a first output indicative of a first context extracted from one or more test logs based at least on one or more first prompts and first information retrieved from one or more first databases, wherein the one or more test logs include a record of automated test execution for one or more tests and at least one test failure; provide the first output indicative of the first context extracted from the one or more test logs, one or more second prompts, and one or more product logs as input to one or more second models, wherein the one or more product logs correspond to the one or more test logs and include a record of at least one event or action that occurred within a product during the one or more tests; generate, using the one or more second models, a second output indicative of a second context extracted from the one or more product logs corresponding to the one or more test logs based at least on the first output indicative of the first context extracted from the one or more test logs, the one or more second prompts, and second information retrieved from one or more second databases; and perform one or more operations for addressing the at least one test failure based at least on the second output indicative of the second context extracted from the one or more product logs.
Facts
- Company
- Nvidia (NVIDIA Corporation)
- Official title
- SYSTEMS AND METHODS FOR ROOT CAUSE FAILURE IDENTIFICATION USING MULTI-STAGE RETRIEVAL-AUGMENTED GENERATION
- Publication number
- US20260310595A1
- Application number
- 19170398
- Filed
- Apr 4, 2025
- Published
- Oct 8, 2026
- Inventors
- 3
- Claims
- 20
Classified under
G06F 11/3692 Physics & computingG06F 11/3684 Physics & computingG06F 11/3688 Physics & computingMore from Nvidia in AI & software
- Nvidia patents AI coach that analyzes your gameplay to suggest strategies Oct 13, 2026
- Nvidia patents AI system that writes and voices character conversations Sep 10, 2026
- Nvidia patents system that uses AI to find bugs in game videos Sep 17, 2026
- Nvidia patents AI method to make digital characters move realistically Sep 8, 2026
- Nvidia patents AI system that builds image generators from text prompts Sep 15, 2026
Summaries are simplified from the filing’s own abstract and first claim. A patent filing is not a product announcement.
Patent data and drawings: U.S. Patent and Trademark Office (USPTO). Nebula is not affiliated with or endorsed by the USPTO.