AI Reality Case Studies

Almost Magic Tech Lab seal: Clarity over cynicism. Less noise, more nerve.

By Almost Magic Tech Lab

AI Reality Case Studies

Fictional composite cases about real patterns in how organisations adopt AI.

What these are

Each case follows one organisation through a pattern that keeps showing up in real AI adoption: a sensible shortcut, pressure to move quickly, and the moment a small compromise quietly becomes the rule. The organisations and people are fictional composites. The patterns are not - they come from Almost Magic Tech Lab's advisory work, anonymised so the lessons travel and the clients do not.

Every case is written to be read in one sitting, in plain language, with the decisions shown as they looked at the time rather than with hindsight.

Why they are free

Because we would rather these mistakes were made less often. Almost Magic Tech Lab advises organisations on AI adoption, and the same avoidable patterns keep repeating across companies and industries. Publishing the patterns free is the fastest way we know to help the market recognise them early.

Free means free. No trial, no tier, no card. Reader editions need no account and no sign-up, and this site runs no analytics and no trackers.

Who they are for

Teams. Read a case together and argue about what the people in it should have done. The argument is the point: it surfaces your own assumptions before they get expensive.

Training and education. Each case maps to a domain - ARC-01-02 covers data quality and analytics - and comes with material built for structured sessions.

Leaders. The cases are short enough to read before a meeting and specific enough to change what happens in it.

Reader editions and facilitator guides

Every case has two editions. The reader edition is open to everyone: read it here, or download the PDF and forward it.

The facilitator guide is for the person running the session: the run sheet, the discussion questions, the moments to slow down, and the quality checks. Guides are sent on request rather than posted publicly, because they work best in the hands of someone preparing to lead a room - and because we like knowing who is teaching with these.

What happens when you ask for a guide

One email. You give us your address and tick the consent box; we send you a link to the guide, valid for 24 hours. That is the whole transaction.

If you also tick the updates box, we will email you when a new case study arrives. Nothing else. If you do not, you will not hear from us again.

About Almost Magic Tech Lab

Almost Magic Tech Lab Pty Ltd is an independent advisory lab. We help organisations make calm, evidence-led decisions about AI: what to adopt, what to avoid, and how to tell the difference. These case studies are one part of that work, alongside advisory engagements and the free book Someone Said AI. Now What?.

Read the cases

ARC-01-02 · Data Quality and Analytics

When one useful data shortcut quietly became the rule

Read the case
ARC-HR-01 · People and Hiring

The shortlist nobody could explain

A hiring shortlist hides a cut line. Who gets a human review before the interview list is locked?

Read the case
Cybersecurity · Cybersecurity and Trust

The incident note that was quicker than the incident

A polished incident note outruns the evidence. What can the board be told, and which data route must pause?

Read the case
Customer Service · Customer Service and Support

The complaint the queue called closed

A polite reply becomes a closed case. Who checks the charge, and what must change before the pilot expands?

Read the case

New cases arrive with each edition. Updated regularly