Checklist
Enterprise AI to production — the checklist
33 questions to put to your own team before enterprise AI leaves the pilot. They are the ones we end up asking in every engagement, in the order the problems tend to appear.
How to use this
Do not answer these yourself. Print it, hand it to the people who run the systems, and ask them to tick only what they could demonstrate this week — not what is planned, not what is nearly done. The ticks are less useful than the gaps: a question nobody can answer without a meeting is the finding.
Four sections, in dependency order. Grounding failures make governance unprovable; governance gaps make orchestration unsafe; and a sovereignty claim you cannot evidence undoes all three the first time somebody asks.
Tick as you go and the panel keeps a running score. It is indicative — a prompt for the conversation, not a maturity rating — and it stays in your browser: nothing is sent to us, and printing takes the score with it.
Score: 0 of 33 (0%)
Grounding 0/8 · Governance 0/10 · Orchestration 0/8 · Sovereignty 0/7
1. Grounding
Whether the answers rest on something the organisation recognises as true.
2. Governance
Whether someone is accountable, and whether the evidence exists to show it.
3. Orchestration
Whether whole processes run, or whether each agent stops at the edge of one team's tools.
4. Sovereignty
Whether the claim you make about control would survive being tested.
If the gaps are real
Bring the unanswered questions to a 30-minute briefing. We go through what your answers imply for grounding, governance and deployment — no slides, and no obligation to do anything afterwards.
Book the 30 minutes belowscrydon.com/contact-us/