You can't adopt AI successfully without this

Your team's Slack now has ChatGPT, Claude, and Copilot integrations. The dev team is using Claude Code. Marketing uses one tool, Engineering uses another, and no one's sure if anyone's actually getting measurable value. You're paying for 5-10 AI subscriptions and can't name a single workflow that's demonstrably better.

Or maybe it's bigger than that. Your CEO just announced an AI transformation initiative. Engineering is spinning up pilots. Procurement is fielding vendor calls. Meanwhile, you're wondering: are we actually ready for any of this?

Here’s what I’ve noticed working with teams at different scales. Most organisations jump straight to solutions because they already work with that vendor. They evaluate tools, run pilots, get a few good demos, then nothing. They skip the crucial first step, which is understanding where they actually are.

You can't build a roadmap if you don't know your starting point.

The Diagnostic Gap

I see this pattern more than you might imagine.

A startup hires its first AI engineer without defining which problems actually need AI versus better processes. An enterprise launches five pilots across different departments with no shared governance. A product team integrates an LLM but hasn’t thought through what happens when it hallucinates in front of a customer.

The result:

• Scattered pilots that never scale

• Governance frameworks that arrive six months too late

• Teams that are not equipped to adopt what you are building

• A growing sense that you are spending money without a clear return

The companies that avoid this mess do something simple. They diagnose before they prescribe. They figure out what is actually missing before they start filling gaps.

The AI Readiness Audit

Over the past few months, I’ve been developing a diagnostic framework I’m calling the AI Readiness Audit. It’s not a checkbox exercise. It’s a structured way to measure where you are across the dimensions that decide whether AI initiatives succeed or quietly stall.

The full audit looks at things like:

• Strategy and use case clarity

• Data, content and governance

• Operating model, ownership and adoption

Coming soon, I’ll share the full framework, how to score yourself, what different maturity levels mean, and specific actions for each gap you find.

Before that, I want to share six questions that will immediately show you whether you need this diagnostic. Think of these as warning lights on the dashboard. If you struggle to answer more than two of them confidently, you have work to do before your next AI investment.

Six Questions That Reveal the Gaps

Question 1: Can you name three use cases where AI would create measurable value in the next 90 days?

Why it matters: If leadership can't get specific, you're operating on hype rather than strategy. "Improve productivity" isn't a use case. "Reduce support ticket resolution time from 24 hours to 4 hours" is.

What good looks like: You have a prioritised list with actual metrics attached. Someone owns each use case. You know what success means.

Question 2: Who owns AI governance in your organisation right now?

Why it matters: If the answer is "everyone" or "we're forming a committee…," no one actually owns it. When your first AI incident happens - and it will - there's no clear decision maker.

What good looks like: One person wakes up thinking about AI risk. They have budget, authority, and a framework for making calls when things go wrong.

Question 3: What happens when an AI system makes a mistake that affects a customer?

Why it matters: This question reveals whether you have incident management processes or just hope. Most organisations discover their gaps during a crisis, which is expensive and embarrassing.

What good looks like: You have a documented process. Your team has run through scenarios. Someone knows how to pull the plug if needed.

Question 4: Can your engineering team explain how your AI systems make decisions?

Why it matters: If your engineers can't explain it, you can't debug it, audit it, or improve it. You're flying blind.

What good looks like: Your team maintains documentation on model behaviour, edge cases, and failure modes. They can walk someone through why the system did what it did.

Question 5: How do you measure whether AI is actually working?

Why it matters: "Users seem to like it" isn't a metric. Without measurement, you can't tell success from theatre. You're making renewal decisions based on vibes.

What good looks like: You track specific outcomes - time saved, errors reduced, revenue influenced. You review metrics regularly and kill things that don't deliver.

Question 6: What's your plan when the tool you've built around gets shut down or changes dramatically?

Why it matters: API deprecations happen. Pricing changes overnight. Models (can) get worse with updates. If you haven't thought through contingencies, you're one vendor decision away from a crisis.

What good looks like: You've documented dependencies. You know what it would take to switch providers. Your contracts have some protection built in.

What These Questions Reveal

If you struggled to answer more than two of these confidently, you're not alone. Many organisations have not yet fully defined processes robust enough to manage the change of safely introducing AI, whether they're a three-person startup or a three-thousand-person enterprise.

The difference between successful AI adoption and expensive experiments comes down to honest diagnosis. Knowing what you don't know. Building the foundation before you scale.

The six questions in this issue are the surface layer of the AI Readiness Audit. The full framework goes deeper and turns these signals into a simple score and a concrete starting plan.

Coming soon, I'll share the complete AI Readiness Audit framework - how to score yourself across six dimensions, what each maturity level means, and the specific actions that move you forward.

I’m building atomictheory.ai as a home for the audit and for my work helping organisations navigate AI transformation. The goal is simple, give you a ten minute diagnostic, a clear readiness score and a focused next step, not another 80 page slide deck!

Until then: Reply and tell me which question hit hardest. Where does your organisation have the biggest gap?

The Atomic Builder is written by Faisal Shariff and powered by Atomic Theory Consulting Ltd — helping organisations put AI transformation into practice.