Field notes from two decades of building software and companies.
An agent can do the work without bearing the consequences. Before granting production authority, name who owns the outcome, exceptions, approvals, and recovery.
Read article
Replace soft time-saved claims with a workflow baseline, quality floor, decision rule, value-capture mechanism, and named owner finance can inspect today.
Read article
Fund agent-readable infrastructure around one bounded workflow with explicit capabilities, permissions, actions, receipts, ownership, and operating proof.
Read article
Stop expanding AI seats until one workflow shows fewer manual touches, shorter cycle time, less rework, controlled exceptions, and lower cost-to-serve.
Read article
Five filters for choosing the first workflow to automate.
Read article
CFOs should require five answers before approving AI budgets: measurable outcomes, accountable ownership, full operating costs, rollback, and mapped workflows.
Read article
Leaders should approve AI for production only when a workflow has release evidence for quality, security, evaluator validity, ownership, and reviewer capacity.
Read article
Build an AI strategy around agent-ready operations, not a tool list. Map workflows, constraints, permissions, ownership, and rollback paths before you buy.
Read article
A decision guide for authorizing ambient AI work and choosing the minimum trigger, control, and evidence backbone the workflow requires.
Read article
A private AI release packet turns acceptable work into versioned evidence for task quality, security hard fails, evaluator validity, cost, latency, and business outcomes.
Read article
A five-check operating test for deciding whether a workflow needs cleanup, a fixed-scope AI pilot, permanent human review, or no automation.
Read articleGet occasional field notes on choosing models, building useful AI workflows, and making better decisions with the tools.
No fixed schedule. We will write when there is something worth sharing.
Confirm once by email · Unsubscribe anytime.