Security
The controls come before the code.
Delightful Computer builds AI systems for regulated teams in pharma, legal, financial services and other work where a wrong answer is expensive. Security is part of the system design: data boundaries, identity, reviewer control, logging and rollback are specified before production work starts.
Common controls: least-data design, scoped credentials, client-owned deployment, human review paths, reviewer-visible reasoning, prompt and model metadata, retrieval traceability and audit-ready evidence. The point is a system the people who own the risk can inspect.
The public website does not expose customer workspaces, operational write actions or production APIs. Engagement systems are scoped, authenticated and documented inside the client environment. Send security reports to morgan@delightful.computer.