Engineering · · 2 min read
How we keep your data out of public models
AI systems handle sensitive operational data, so security is designed into every layer. Here is what that means in practice, and our honest compliance status.
Our systems handle sensitive operational data: customer records, contracts, tickets, orders. So security is designed into every layer, not added at the end. These are the practices behind every system we build.
Your data never trains public models
Model providers are contracted not to retain or train on your data. Before any request reaches a model provider, personal data is removed. The model sees what it needs to do the job, and nothing more.
The practices
- Encryption: AES-256 at rest and TLS in transit, everywhere.
- Least privilege: strict role-based access and isolated networks for every environment.
- Data minimisation: personal data is removed before any request reaches a model provider.
- Zero retention: model providers are contracted not to retain or train on your data.
- Secrets management: credentials live in managed vaults, never in code.
- Monitoring: continuous monitoring and alerting across every deployed system.
Security is part of the architecture, not a checklist at the end.
Our honest status
We build to SOC 2 standards. We are not yet certified, and we will not pretend otherwise. If you need detail for a vendor review, ask us for our current controls documentation.
The full picture is on our Security page.
Written by 121Labs
Next entry · Automation
Start where the work piles up