When I finished my Data Analytics certificate, I faced a challenge that a lot of people hit after completing a course: taking polished theory and applying it to a real, open-ended project with no guardrails and no answers at the back of the book.
The Bellabeat Case Study became that challenge. And I used AI as a project partner throughout — not to generate my analysis, but to stress-test it, sharpen it, and make sure what I'd found actually came across in the final report. Here's an honest account of how that worked.
The analysis itself — every line of code, every visualisation, every data cleaning decision — was mine. What AI helped with was the layer on top: making sure the insights I'd found were communicated as clearly and persuasively as the data deserved.
Three roles AI played — and what I actually asked it to do
After completing my analysis, I needed to know if my argument was strong enough for a professional audience. I used AI as a reviewer — not to tell me what to think, but to tell me whether what I thought was coming across.
The most powerful help came in turning good analysis into a great report structure. My data showed two distinct user groups — a Majority Audience and a High-Risk segment — but I was treating them as separate findings rather than a connected argument.
Before finalising, I had a structural question: was it okay for my Key Findings section to contain mini-recommendations when my Executive Summary also had big-picture recommendations? I wasn't sure if that created confusion or useful layering.
The principle behind all of it
Looking back, there's a clean logic to how AI contributed at each stage — and a clear boundary it never crossed.
Why this matters beyond one project
I tell this story because it's a concrete example of the kind of AI use that actually delivers value — and that's often missing from the conversation. The debate tends to swing between "AI wrote everything" and "AI is cheating." Neither captures what's actually useful.
What's useful is treating AI the way you'd treat a very capable colleague who has no context about your work, your client, or your goals — but who can give you honest, structured feedback the moment you bring them up to speed. That framing changes what you ask for, how you interpret the response, and what you do with it.
The output quality in the Bellabeat project was higher because I used AI at the right stage, for the right job, and kept the actual thinking where it belonged: with me. That's not a compromise. That's just good collaboration — with a new kind of partner.