Articles

Operator notes.

Problem-first writing on AI, automation, and the process underneath. Each piece names a real bottleneck, exposes the broken layer nobody owns, and shows what the working system actually looks like.

Enterprise AI

AI agent reliability is a systems problem, not a model problem

AI agent reliability isn't about a flakier model. Ookla's 2026 report shows the real shift: an agent is a dependency chain across systems you don't monitor end to end.

7 min read →
Conversational commerce

Meta Business Agent can close a sale in a chat. Whether it should is a data problem.

Meta Business Agent now books appointments and closes sales on WhatsApp. The agent isn't the hard part — whether your inventory and calendar data can back the promise it makes is.

7 min read →
Enterprise AI

AI agent costs just went metered. The problem is nobody owns the meter.

AI agent costs flipped from flat per-seat licenses to metered per-token bills in 2026. The real failure isn't the price — it's that no one in the org owns the meter.

8 min read →
Construction

AI contract review for construction is the cheap part — the obligations are the problem

AI contract review for construction can redline a deal in 24 hours. But construction disputes come from obligations no one tracks after signing, not from bad clauses.

6 min read →
Field service

Autonomous field service dispatch is only as good as your data

Field service AI went from assisted to autonomous dispatch in 2026. But autonomous scheduling commits to whatever your field data already is — at machine speed.

6 min read →
Enterprise AI

Integration and data contracts: the layer that decides whether AI agents work

The 2026 protocol wave (MCP, A2A) connected agents to your systems. It didn't make your systems agree. Why integration and data contracts decide whether enterprise AI works.

5 min read →