Insights
Perspectives, not press releases.
What we believe about enterprise AI, written for the people accountable when it goes wrong.
THE PERSPECTIVE · PDF
The Lights-Out Enterprise
The full thesis as a shareable four-page Perspective — for your board pack, not your browser tab.
Download →The Sentient AI thesis · July 2026
The Lights-Out Enterprise
Our defining perspective: Lights-Out is an operating model decision, engineered in four moves — TOM, Data Model, governed agents, evidence. With the download-ready PDF.
Read the thesis →The operating model · July 2026
Agents take the intelligence. People keep the judgment.
The principled line inside every function — and why managing it as a ratchet, on evidence, is what makes autonomy compound.
Read →Governance · July 2026
The evidence dividend
Why governed autonomous operations audit better than the human processes they replaced — populations over samples, verification over trust.
Read →The playbook · July 2026
The first ninety days of Lights-Out
Monday morning, what actually happens: the wedge, the operating model, the autonomy ramp — and a measured delta by day 90.
Read →The thesis · July 2026
Don’t buy the tool. Buy the work.
Services are the new software. Why Sentient.AI Agentic sells finished, evidenced work — and how TOM + Data Model gets you to Lights-Out.
Read →The thesis · June 2026
Lights-Out is an operating model, not a feature
Why "AI adoption" plateaus at copilots — and what separates enterprises that assist work from enterprises that hand work over.
Read →The framework · June 2026
Why the enterprise AI stack has exactly five layers
Most AI capability maps mix techniques, lifecycles, and controls on one axis. A MECE stack changes what you invest in first.
Read →The CFO office · June 2026
Lights-Out Finance: what the CFO owns when agents run the close
Autonomous finance doesn’t shrink the CFO’s job — it moves it up to judgment, exceptions, and the control plane.
Read →Governance · June 2026
Guardrails before autonomy: the license for the agentic enterprise
Control-first organizations scale agents faster. What real guardrails are, and a three-question test for yours.
Read →Operations · June 2026
Why your AI pilots stall: the operations layer nobody budgets
Pilots die between demo and production because operations is a capability, not a project. The platform math that fixes it.
Read →Operating model · June 2026
From shared services to self-running services
Labor arbitrage bought a discount on manual work. Autonomy removes the work — and the ceiling that came with it.
Read →Perspectives by email