Published Analysis

The Autonomy Boundary

An examination of where agent authorization ends and human accountability begins, and of the seven runtime controls that make that boundary auditable. This series produced the Autonomy Boundary Framework.

• Part 1The problem: capability is not permission, and the distinction is rarely documented.
• Part 2The controls: conditions required before an action crosses the boundary, at the crossing, and afterward.
• Part 3Legibility: how approval and authorization diverge in practice, and the operational cost when they do.

Read the series →

The Last Mile

An analysis of AI deployment in regulated industries. Each sector examined presents functioning models and stalled production deployments; the series examines the regulatory and operational gap between the two.

• Week 1Finance — model performance versus audit-trail requirements in production approval.
• Week 2Healthcare — HIPAA, clinical risk, and error cost in high-confidence systems.
• Week 3Retail — scale, latency, and the economics of marginal error at volume.
• Week 4Government — FedRAMP and NIST 800-53 as architectural design inputs.

Read the series →

Ongoing Publication

New analysis publishes on LinkedIn, where it is reviewed in public by practitioners operating these systems. Several open design discussions in the framework repository originated in that review.

View on LinkedIn →