Self-check · AI in production

The Enterprise AI Readiness Self-Check

Twelve questions across the five dimensions that decide whether your organisation can move AI from pilot to production — strategy, data, talent, infrastructure, and governance.

June 20, 20264 min readInteractive

Take the self-check

Most enterprise AI never ships. It demos well, gets a budget line, and stalls before the first real user.

Twelve questions tell you whether yours will make it — and which dimension will stop it.

Questions: 12
Dimensions scored: 5
To answer: 3 min
Answers stay in your browser: 100%

How it works

Answer honestly
Work top to bottom. A “not sure” is an answer too — it points at the thing nobody owns yet.
Your answers stay with you
Answers are saved only in this browser — no account, nothing to submit.
Take it into the room
Download the PDF and bring the same questions to the meeting where the decision actually gets made.

At a glance

Format: Self-check
Topic: AI in production
Published: June 20, 2026
Reading time: 4 min read

AIEnterpriseReadiness

Download PDF

Get in touch

The pattern is consistent enough to predict: the organisations that cross into production share a named executive sponsor, a success metric tied to a business outcome, and at least one internal champion. The ones that don't, don't.

This is a pragmatic self-check. Answer each question honestly with yes, no, or not sure — a "not sure" counts as a no. Twelve questions, five dimensions. Your score builds as you answer, and your answers stay in your browser.

0 of 12 answered

Strategy and sponsorship

Is there a named executive sponsor who owns the outcome — not just the budget?

A sponsor who can unblock a data-access fight or a procurement freeze is worth more than the model.

Is the first use case tied to a metric someone already reports on?

"Reduce handle time," "recover abandoned carts," "cut Rx leakage" beat "explore AI."

Can you name the internal champion who will use the thing every day?

Pilots without a daily user are demos.

Data

Do you know where the data lives, who owns it, and whether you're allowed to use it for this?

Data readiness is the single most-cited cause of AI failure — access rights and data lineage block pilots more often than data volume.

Is the data good enough to be wrong about in production?

You need documented failure modes and edge cases before customers discover them for you.

Talent and literacy

Is there someone on the team who can tell a good output from a plausible-but-wrong one?

Domain experts who evaluate model outputs daily matter more than generic accuracy dashboards.

Do the people around the system — support, ops, compliance — understand what it can and can't do?

Operations teams need explicit playbooks for model hallucinations before the first customer ticket arrives.

Infrastructure and integration

Can the system reach the systems it needs to — read and write, by role?

Production requires writing records back to CRM/ERP systems, triggering workflows, and enforcing RBAC.

Is there a path to monitor it in production — latency, cost, drift, and a way to roll back?

If you can't see it degrade, you can't keep it alive.

Governance and the EU AI Act

Have you classified this system under the EU AI Act?

The four risk tiers decide your obligations. The Act applies since August 2026; high-risk duties land 2 December 2027 (Annex III) — classification now is what makes that date cheap.

Is there human oversight built into the core architecture?

Someone must be able to understand, override, and detect anomalies — by design.

Do you have one accountable owner for AI governance?

A named individual, rather than a quarterly committee.

Where you scored low is where an embedded product lead earns their keep — turning a stalled pilot into a measured outcome is most of what I do. If three of these questions made you wince, that's usually the conversation worth having.

Related resources

Quick-check

The EU AI Act Product Quick-Check

The EU AI Act applies since August 2026, and the high-risk obligations land on 2 December 2027. Place every AI feature in its risk tier — the four tiers, the Annex III trigger areas, and the five things a product team must have in place now.

5 min readInteractive

Run the quick-check

Guide

The Enterprise Product Operating Model Guide

How to run product inside a DACH enterprise without drowning in governance — team topology, decision cadence, stakeholder alignment, and where an embedded product lead plugs in.

5 min readInteractive

Read the guide

The conversation after

Thirty minutes, no deck. Bring what you just answered and we work out what it means for your platform.