Normalized claim
Quantified impact: 50% increase
The automatic correction rate improved from about 50 percent to about 80 percent with the o1 model.
Woven by Toyota: Multi-agent Azure OpenAI workflow to auto-fix MISRA compliance errors
Use case typeLegal workflow automationUpdated Jun 13, 2026
Woven by Toyota, part of the Toyota Group, used Azure OpenAI Service to automate MISRA code-compliance fixes for embedded C/C++ software in autonomous driving and ADAS development. The team built a multi-agent workflow with Coder, Reviewer, and Evaluator agents to generate fixes, review them, and provide reasoning and certainty for engineers. The implementation integrated Azure App Service, Azure Cosmos DB, AutoGen, and GitHub Enterprise CI/CD, and was tested on sample code and in-house code.
- Organization
- Toyota
- Industry
- Automotive
- Location
- Japan
- Published
- January 2025
Planned next steps
- The source says the organization aims to achieve Accuracy: 81.5%.
- In a proof-of-concept on in-house code, the system achieved a 97.1 percent code generation success rate.
Why do we believe this?Outcome claims, sources, and evidence checks
Normalized claim
Quantified impact: 80% increase
The automatic correction rate improved from about 50 percent to about 80 percent with the o1 model.
Normalized claim
Quantified impact: 97.1%
In a proof-of-concept on in-house code, the system achieved a 97.1 percent code generation success rate.
Normalized claim
Accuracy: 81.5%
The project automatically corrected 81.5 percent of MISRA compliance errors.
Why do we believe this deployment?Customer identity, provider attribution, maturity, and source checks
- Customer
- Toyota
- Provider
- Microsoft
- Maturity
- Unknown
- Linked source
- Microsoft Customer Stories
No explicit deployment-stage evidence found.
Customer identity supportedSource describes one deploymentMaturity evidence evaluated
Primary read
Use case focus
Showing 3 of 3
- 1Code compliance automation
- 2Software development automation
- 3Agentic workflow
- Woven by Toyota first tested Azure OpenAI Service GPT-4o to automatically fix MISRA compliance errors in sample and in-house code.
- The project evolved into a multi-agent system with a Coder agent that fixes code, a Reviewer agent that critiques and iterates on the corrections, and an Evaluator agent that produces reasons and certainty factors.
- The solution used Azure OpenAI Service reasoning and coding models, AutoGen for orchestration, Azure App Service, Azure Cosmos DB, and GitHub Enterprise integration to support development workflows.
Architecture
The article describes a multi-agent architecture in which a Coder agent proposes MISRA-compliant fixes, a Reviewer agent iteratively checks and refines them, and an Evaluator agent assesses final outputs with reasoning and certainty factors. The system uses AutoGen for orchestration and integrates Azure OpenAI Service, Azure App Service, Azure Cosmos DB, and GitHub Enterprise/CI-CD.
Sources & evidence1
Evidence: Medium50/100Evidence strength
- Customer explicitly identified
- Primary source available
- Quantified outcome available
- Technical implementation details available
ExpandedExpanded
The same organization appears in newer AI deployment evidence.
- Same organization re-documented as recently as 2026.
Measures whether this deployment's public evidence persists — not whether the system is still in production.
Type: Customer StoryPublished: Jan 30, 2025Publisher: MicrosoftEvidence: PrimaryConfidence: High
AI-generated summary. Verify important details with the linked sources before relying on this case.
Explore related AI use cases
Customer evidence
Provider evidence
AI patterns
Browse the catalog
Was this useful?
Community
Comments
No published comments yet.