The Real AI Readiness Assessment: 10 Questions That Actually Matter

Most AI Readiness Assessments Are Vendor Qualification Forms

“Do you have a data warehouse?” “Is your leadership aligned on AI?” “Have you identified AI use cases?” These are the questions most AI readiness assessments ask. They sound strategic. They’re actually qualification questions designed to figure out if you’re ready to buy something.

We’ve run enough AI strategy engagements to know which questions actually predict whether AI will work for an organization — and most of them have nothing to do with technology.

The 10 Questions

1. Can you name the specific task that’s burning the most time?

Not “we want to use AI for efficiency.” A specific task. “Our loan analysts spend 3 hours per application re-keying data from tax returns.” That’s a starting point. “We want to leverage AI across the organization” is a budget request, not a project.

2. Is the person doing that task involved in this conversation?

AI projects that start in the C-suite and get pushed down fail at 3x the rate of projects that start with the person doing the work. The analyst knows which fields are wrong most often. The consultant knows which prep steps matter most. Strategy sets direction, but the subject matter expert defines the product.

3. What does “good enough” look like?

AI doesn’t need to be perfect. It needs to be better than the current process. If your current process has a 5% error rate and takes 4 hours, AI with a 3% error rate that takes 20 minutes is a massive win. But if you’re expecting zero errors, you’re going to be disappointed with every AI product on Earth.

4. Do you have examples of the inputs and outputs?

Not schemas. Not data models. Actual examples. “Here’s a real tax return. Here’s the spreadsheet our analyst fills out from it.” If you can’t produce 10 examples of input/output pairs, you’re not ready to build an AI product — you’re ready to document your process.

5. Who will validate the AI’s output?

Every production AI system needs a human in the loop. Not because the AI can’t do the work, but because someone needs to catch the cases where it doesn’t. If you don’t have a clear answer to “who reviews this before it goes to the client/customer/regulator,” you’re not ready.

6. What happens when the AI is wrong?

This is the question most organizations skip. If the AI extracts the wrong number from a tax return, what’s the blast radius? A minor correction? A compliance violation? A wrong loan decision? The answer determines how much validation infrastructure you need to build.

7. Can you afford to be wrong publicly?

Customer-facing AI (chatbots, recommendation engines, content generation) can embarrass you in ways internal tools can’t. If your brand can’t survive the AI saying something stupid to a customer, start with internal tools.

8. Do you have someone who can own this for 6 months?

AI products aren’t fire-and-forget. They need feedback loops, iteration, and someone who cares about whether they’re working. If nobody in your organization will check on this weekly for 6 months after launch, it will quietly fail.

9. What are you going to stop doing?

AI should replace work, not add to it. If your answer is “nothing — we’ll add AI on top of what we already do,” you’re increasing complexity without reducing cost. The ROI comes from eliminating steps, not layering new ones.

10. Have you budgeted for the boring parts?

The AI model is the exciting part. Data cleaning, integration, testing, training, change management, and ongoing maintenance are the boring parts. They’re also 70% of the total cost. If your budget covers the model and nothing else, multiply your estimate by 3.