Building DiscernAI: Decision-Making for Humans, Not Algorithms

I built DiscernAI because I was frustrated with my own decision-making. Not because I lacked information — I had too much of it. I’d open twelve tabs, read conflicting takes, and still feel stuck. So I did what engineers do: I decided to build my way out of the problem.

That was my first mistake.

The real problem wasn’t what I thought it was

My original assumption was that people needed better structure for decisions. A framework. Criteria, weights, a score. Clean and rational.

I built exactly that. And when I handed it to early users, something unexpected happened: they’d go through the whole flow, see the recommendation, and then… do whatever they were already going to do anyway.

That stung. But it taught me something more interesting than I’d bargained for — people don’t actually want a better answer. They want confidence in their own answer. The decision was usually already there, half-formed. What was missing was permission to trust it.

So I stopped designing for optimization and started designing for clarity. The AI’s job isn’t to decide — it’s to surface what you already value, make the tradeoffs visible, and get out of the way.

What I’d do differently

I spent the first two months building the scoring engine before I’d talked to a single real user about a real decision they were stuck on. Classic mistake. When I finally did those conversations, almost every design assumption I’d made turned out to be slightly off — not catastrophically wrong, just tilted toward how I make decisions, not how most people do.

If I started over: five user interviews before a single line of code.

The tech stack — SwiftUI, Gemini AI, Supabase, CoreData — was mostly the right call, but I underestimated how much of the product experience lives in the framing of the AI prompt, not the infrastructure around it. I probably spent 60% of my time on architecture and 40% on what actually moved the needle for users.

What surprised me

350+ people signed up without a single paid ad. That told me the problem was real — people are genuinely exhausted by the volume of choices and the noise around them. But the use cases surprised me. I expected big life decisions. What I got was: which streaming show should I watch tonight, which dog food is healthiest, should I take this job or that one.

The mundane decisions and the major ones turned out to need the same thing: a way to get out of your own head and see what you actually care about.

Why it matters to me

I’m a PM by trade, which means I spend a lot of time at the intersection of what’s technically possible and what users actually need. DiscernAI was the first time I owned that entire surface — the product decision, the design, the engineering, the positioning. It made me a better product thinker than six years of working on other people’s roadmaps did.

I built it because I believe tools should enhance human judgment, not replace it. In a world that increasingly wants to decide things for us, staying the author of your own choices feels like something worth protecting.

Try it on the App Store →