AI models and providers
Windmill uses state-of-the-art AI models from trusted providers: These models power Windy, our AI agent, to help with feedback collection, recap generation, and performance insights. Zero data retention Windmill has zero data retention enabled across all AI model providers. This means:- Your data is not stored by our AI providers after processing
 - API calls are processed and immediately discarded
 - No logs or caches of your data are retained by OpenAI, Anthropic, or Google
 
Zero data retention ensures that your company’s data never becomes part of AI provider training datasets or storage systems.
Data retention
Windmill retains data from customers as long as they have an active engagement with Windmill. Customers have the right to request complete data deletion after ending an engagement.Telemetry
To enhance the performance and accuracy of our AI models, Windmill collects telemetry data, encompassing inputs and outputs of AI interactions. Access to this data is strictly limited to authorized personnel, underpinned by robust security protocols and comprehensive audit logs, safeguarding customer information while facilitating continuous improvement of our AI solutions.Model training
We do not train public models on your data. Here’s what that means:- Windmill does not build or train generalized AI models from customer data
 - Any custom AI models we train are for internal use only, strictly to operate and enhance our product
 - We never train models to be made available as a standalone service or via an external API
 - When we train custom models, they’re exclusively derived from generated synthetic datasets
 - Our vendor partners (OpenAI, Anthropic, Google) do not use logs to train their own models
 - With zero data retention enabled, your data never persists in vendor systems
 
Your data stays your data. We never use your company’s information to improve models that serve other customers.
Prompt engineering
The optimization of AI performance is a continuous endeavor at Windmill. Our team regularly examines telemetry logs to refine and improve AI prompts, enhancing the relevance and accuracy of AI-generated content. This process is conducted with the utmost respect for privacy—no private customer data is incorporated into prompts or used in any unauthorized manner.Model accuracy and quality
Generative AI, while powerful, can sometimes yield unexpected outcomes. Windmill is committed to mitigating instances of low-quality, incoherent, incomplete, or insensitive content. We employ rigorous quality control measures and are dedicated to iterative enhancements of our AI capabilities. If you encounter any issues with AI-generated content, please report them to support@gowindmill.com.AI attribution in reports
Windmill attributes engineering work and contributions based on activity in connected systems like GitHub. Commit attribution Commits are attributed based on the author of the commit, which is configured in local git configuration (not GitHub user identity). To ensure commits are attributed correctly, configure your local git:- Devin
 - Claude Code (via GitHub Actions)
 
- OpenAI Codex
 - Cursor
 - Claude Code (locally)
 
If you notice incorrect attribution in your Recaps or reports, check your git configuration and coding agent settings.