Building infrastructure for production AI agent systems. We're focused on the gap between prototype and production: helping engineering teams build agents that are reliable, verifiable, and cost-efficient in the real world.
- deconvolute — Runtime policy enforcement for MCP tool calls using CEL-based rules.
- legend — PII pseudonymization for the full agentic loop. Intercepts at all four boundaries; restores originals in the final response.
- pii-benchmark — Reproducible head-to-head benchmark for PII detection and pseudonymization. Currently compares Legend and Presidio across eleven entity types.
Follow along or reach out if you're working on production agent systems.
