Enterprise LLM proxy that strips PII before it reaches the model. Self-host for free or use Veil Cloud. Multi-provider, OpenAI-compatible gateway.