Insights
Thinking on agent accountability
Practical writing for the people who own AI agents in production: how to measure what agents produce, how to trust what they report, and what regulators will expect you to show.
- Metrics
Cost per outcome, straight-through and takeover rates, and how to compare agents fairly.
- Evidence
Reconciling what agents claim with what systems of record actually show.
- Regulation
The EU AI Act, DORA and what agent owners need to be able to show.
All articles
3 articles · Metrics, Evidence, Regulation
Metrics
Cost per outcome: the metric that turns agent spend into a decision
Token bills and seat counts tell you what agents consumed. Cost per completed workflow tells you what they produced — and whether to scale, reroute or retire them.
November 4, 2026 · 6 min read
Evidence
Two-sided evidence: why an agent’s log is a claim, not a record
Agent runtimes record what an agent attempted. Systems of record record what changed. Accountability starts when you reconcile the two.
November 4, 2026 · 6 min read
Regulation
EU AI Act and DORA: what agent owners should have in place by December 2027
Most enterprise agents won’t be high-risk under the AI Act, but some will — and for EU financial entities, DORA already applies. Here’s what to put in place now.
November 4, 2026 · 6 min read
See it on your own workflow
Reading is one thing. Seeing cost per outcome and two-sided evidence on a real workflow is another.
In a 30-minute walkthrough we reconstruct a workflow end to end — mandates, evidence and cost per completed workflow — and show what a six-week, read-only pilot would cover.
Book a Demo
See your agents through one plane.
Book a 30-minute walkthrough. We'll reconstruct a real workflow end to end — mandates, evidence and cost per outcome — and scope a six-week, read-only pilot around your agents.