Notes on governing intelligence.
Field reports, technical deep-dives, and policy commentary from the team building enterprise AI controls.
Why the AI industry is moving to the token economy
OpenAI, Anthropic, and every major model provider now price AI by the token. As that becomes the industry standard, the token economy is reshaping how the enterprise budgets, scales, and secures AI - here is what is driving it and how to stay in control.
Engineering·Jun 28, 2026DataStrict joins the first wave of Google WebMCP
Browsers are getting AI agents that can act on a user's behalf. We adopted the new WebMCP standard from Google early - and gated every agent action behind a human, by design.
- Engineering·Jun 5, 2026
Designing Edict: a policy language for AI
Governance that lives in a document can't be enforced. Here's how we designed a declarative language so a control is the runtime, not a recommendation.
- Research·May 22, 2026
Adjudication: why one model isn't enough
Sending every request to an LLM-judge is slow, expensive, and non-deterministic. A layered pipeline resolves most requests without a model at all.
- Engineering·May 8, 2026
Shadow mode for policy: simulate before you enforce
Shipping a new control to production blind is how you take down a customer workflow. Replay real and synthetic traffic against a candidate policy first.
Governance·Apr 24, 2026Why AI needs a control layer, not another dashboard
Enterprises don't have an AI observability problem. They have a runtime enforcement problem. Here's what changes when governance becomes infrastructure.
- Security·Apr 11, 2026
Anatomy of an LLM data leak
How a customer support agent quietly egressed PII across three subsystems - and the four enforcement points that would have stopped it.