AI Product Discovery Needs a Process, Not More Tools

The gap isn't tooling — it's structure

Teresa Torres writes in *Continuous Discovery Habits*: "Discovery is not a phase. It's a continuous, structured practice." That word — structured — is the one most teams skip.

What I typically see is a team that runs interviews when there's a question on the table, synthesises loosely, and then lets the output drift into a slide deck or a Confluence page that nobody opens again. When AI enters that picture, it speeds up the transcription and maybe clusters the themes. But the structural problem remains: there's no shared definition of what counts as a validated opportunity, no consistent link between what was learned and what gets prioritised, and no decision log anyone can trace back to.

Engineers get looped in after discovery is "done," which usually means after the decisions have already been made. That's not an AI problem. It's a process problem.

A four-stage discovery loop you can actually run

The teams where I've seen AI genuinely improve discovery all had one thing in common: a repeatable loop with defined stages. AI then accelerates each stage rather than replacing the rigour that was never there.

The four stages are signal collection, synthesis, opportunity framing, and decision linkage.

**Signal collection** is where AI earns its place most clearly. Automated moderation tools like Maze, interview transcription, and survey synthesis can dramatically reduce the time between talking to users and having usable data. But the inputs still have to be structured. If your interview guide is loose, AI gives you a beautifully organised version of loose.

**Synthesis** is where judgment matters most and where most teams let AI do too much. Clustering themes is something AI does well. Deciding which cluster points to a real unmet need — and which is just noise from three unusually vocal users — is something a PM has to own. The interpretive step cannot be automated. My experience is that the teams who forget this end up with a clean-looking opportunity landscape that nobody actually trusts.

**Opportunity framing** is the stage most often skipped entirely. This is where a raw insight ("users are confused during onboarding") becomes a structured opportunity ("first-time users with no admin background struggle to connect their data source without support, which delays time-to-value by an average of six days"). The difference matters. You cannot build a good solution brief off the first version. You can off the second.

**Decision linkage** is what closes the loop — and what most teams never do. Every discovery session should end with one question answered: what do we now believe, and does it change any existing priority? If the answer doesn't connect to something on the roadmap or the opportunity solution tree, the session produced learning without consequence.

The opportunity solution tree isn't a workshop output

If you're not familiar with the opportunity solution tree, it's a framing from Teresa Torres that maps desired outcomes to opportunities to solutions to experiments. What matters here is one thing: it only works as a living artefact.

I've seen teams run a full OST workshop, take a photo of the whiteboard, and never open it again. That's a one-off activity, not a discovery practice. The OST has to be updated after every significant discovery session. It's the connective tissue between what you're learning and what you're building. Without it, discovery and roadmap run as parallel tracks that occasionally collide.

AI can help maintain it — summarising session outputs in OST-compatible language, flagging when new signals challenge an existing opportunity node. But someone has to own it. Assign that to a person, not a tool.

What this looks like in practice

A lightweight version of this operating model for a small product team looks like this: one 60-minute discovery sync per week, a shared decision log with a three-column structure (what we learned, what we now believe, what changes), and the OST reviewed and updated before the session closes.

That's it. No new tools required to start. The AI acceleration comes once the loop is stable.

The question I'd leave you with is a practical one: could your team, right now, answer what your current top opportunity is and show the discovery that supports it?

If not, start there. Build the loop first. The AI will have something real to work with.