Reality Check

Is Your AI Project Theatre?

A brutally honest 11-point scorecard. No sign-up, no fluff, no consulting jargon. If you score badly, at least you'll know before the board does.

Your score: 0/33

Answer the questions below to find out if you're building something real or just keeping the cloud providers happy.

1.

Can you describe the business outcome in one sentence?

Not "AI-driven insights platform." An actual outcome. "Reduce customer churn by 5%." "Cut manual review time by 60%." If the sentence contains the word "leverage," start again.

2.

Have you done the ROI maths honestly?

Platform cost vs measurable value generated. If the answer is "it's strategic" or "we can't put a number on it yet," that's a no. The cloud providers have definitely done their ROI maths. On you.

3.

Is there a customer who cares?

Internal or external — someone who will pay more, stay longer, or do something differently because of what you're building. "The board wants it" doesn't count. The board wanted blockchain in 2017.

4.

Do you know where your AI will be wrong?

AI is a model of reality — not reality. There's always a gap. Can you describe specifically where your system will hallucinate, fail, or give unreliable results? If the answer is "it'll just work" — you're buying magic beans.

5.

Is your data actually any good?

Documented lineage. Known quality. Governed schema. If your ML model is trained on data that was cleaned by someone who left the company, you don't have a data asset. You have a liability.

6.

Who owns production?

If the answer is "the platform team" or "the SRE team" and the people who built it went home at 5pm — you've got ops in a frock, not shared ownership.

7.

Could you ship without your current cloud provider?

Not "would you" — could you? If every component is a proprietary managed service with no abstraction, you don't have an architecture. You have a vendor's reference implementation with your logo on it.

8.

Do your developers understand why, or just what?

If every ticket needs step-by-step acceptance criteria for anyone to know what "done" looks like, the knowledge lives in the process, not the people. Congratulations — you've built the perfect training dataset for AI to replace your team.

9.

Have you killed a project that wasn't working?

In the last 12 months. Not "pivoted." Not "reprioritised." Killed. Stopped. Turned off. Freed up the people and the budget. If not, you're carrying zombie projects and everybody knows it except the sponsor.

10.

Could AI replace your stand-up?

If your daily meeting is just status reporting on small tasks — "I did this ticket, I'll do that ticket, no blockers" — it doesn't need humans. If nobody would notice the difference, the meeting is theatre and so is the process around it.

11.

Can someone in the room say "no" to the HIPPO?

The Highest Paid Person's Opinion isn't always wrong. But if nobody can push back with "cool idea, but here's how that actually works" — you're not doing engineering. You're doing whatever the boss saw at the last conference.

Scored lower than you hoped?

Don't worry — most organisations do. The good news is these are all fixable. Let's talk about which ones matter most for your situation.