"Tell Me About Yesterday Morning" — The Question That Changes an AI Audit

Job Descriptions Lie. By Design, Not by Dishonesty.

In an AI audit, when you ask someone "what do you do here?", they describe an idealized version of their role. Clean. Logical. Fiction.

It's not bad faith. It's how our brains reconstruct a role: we describe what the job should be, not what it actually is day to day.

The Problem: You Map a Process That Doesn't Exist

When an audit rests on that idealized version, everything downstream is fragile:

• You identify AI opportunities on phantom tasks
• You quantify ROI using reconstructed durations, not measured ones
• You present a business case that collapses the moment you move to scoping

Classic outcome: the audit ends up in a drawer.

The Technique That Changes Everything: "Tell Me About Yesterday Morning"

Not the job. Not the theoretical process. Not a typical week.

Yesterday morning. Minute by minute.

The time constraint changes everything. Nobody can reconstruct a clean narrative across 3 specific hours. So the person describes what they actually did — not what they're supposed to do.

What This Question Surfaces

• The real tasks, not the ones on the job description
• Invisible workarounds: the personal Excel file nobody knows about, the copy-paste loop repeated 40 times between two tools, the duplicate email sent because the system doesn't notify anyone
• The frustrations, which are exactly the most profitable AI improvement areas
• Real durations × fully-loaded salary = a quantified cost, not a guess

That's where AI becomes a business case, not a tech fad.

From Tech Fad to Defensible Business Case

An audit built on "yesterday morning" produces numbers you can defend in front of an executive committee:

• "This task consumes 4 hours a week across 3 people, or €55K/year loaded. Automating it costs €18K to build and €4K/year to run. Payback in 5 months."

That can be compared to other investments. Prioritized. Defended.

An audit built on the job description produces: "AI could save time on reporting." No committee makes a decision on that.

Audits That Become Projects vs. Audits That Gather Dust

The difference between an audit that turns into a project and one that stays on the shelf? Rarely the technology. Almost always the interview method.

As long as you map the described job, you're building on sand. The "yesterday morning" question forces you to map the executed job.

A small detail. That changes everything.

---

If you have a team whose real AI gains you'd like to quantify, take 30 minutes to talk it through. We can start with a single "yesterday morning" task.