How to Redesign Your Discovery Process Around AI

The scaffolding has to come first

Teresa Torres put it plainly in *Continuous Discovery Habits*: "The opportunity solution tree is a simple but powerful visual that helps product trios simultaneously manage the business outcomes they've been asked to drive and the customer needs, pain points, and desires that represent potential opportunities for creating value for both the customer and the business."

That structure is the thing AI cannot give you. Without it, AI-assisted discovery produces faster interviews about the wrong problems, more synthesis of conversations that were never connected to an outcome, and opportunity maps that look thorough but have no strategic anchor.

My experience is that teams who feel like AI discovery is shallow have usually skipped this step. They started with the tools instead of the tree.

Before any AI enters your process, you need one clear business outcome your team is working toward, a mapped set of opportunities sitting beneath it, and agreement on which opportunities are worth researching right now. That is the scaffolding. Everything else attaches to it.

There is a sequence, and it matters

The teams I have worked with that genuinely cut discovery time without cutting quality followed a clear order: outcome first, then opportunity mapping, then AI-assisted research and synthesis. Skipping the sequence is why most AI-assisted discovery feels shallow.

Here is what that looks like in practice. The PM defines the outcome — not a feature, not a theme, a specific business outcome the team is accountable for. The team maps the opportunity space beneath it, starting rough and refining as they learn. Then, and only then, does AI come in to help move faster inside that structure.

At the research stage, AI earns its place. Interview moderation, transcript synthesis, pattern clustering across twenty conversations, first-draft opportunity maps built from raw data. These are tasks where AI genuinely compresses time without compressing judgment. A team that previously needed two weeks to go from twelve customer conversations to a synthesized opportunity view can now do it in two days.

But the outcome selection is not AI's job. The strategic call on which opportunities to pursue is not AI's job. The judgment on whether a pattern in the data is actually meaningful or just loud is not AI's job. Those stay with the PM.

What your operating model needs to define

A discovery operating model built around AI should answer one question clearly: who owns what?

From what I have seen, the division should look roughly like this. AI owns the mechanical and the combinatorial: moderation, transcription, synthesis, clustering, first-draft framing of opportunities. The PM owns the directional and the contextual: outcome selection, insight validation, trade-off calls, and the judgment that sits between what the data says and what the business needs.

The artifact that comes out of this is not a quarterly research report. By the time a quarterly report is shared, it is already stale. The output of a well-structured AI discovery process is a living opportunity map that updates as new signals come in — from interviews, from usage data, from support tickets, from whatever sources feed your discovery continuously.

That shift, from document to living map, is the real change. The tools make it possible. The operating model makes it happen.

Where to start

If your discovery feels inconsistent, do not start by adding another tool. Start by writing down the outcome your team is actually working toward right now. One sentence. Then map the top five to eight opportunities you believe sit between you and that outcome.

That takes an afternoon. Once it exists, you have the scaffolding. Then bring in the AI tooling and attach it to something real.

The discovery process gets better when the structure comes first and the tools serve it — not the other way around.