A friend and colleague recently told me about an app idea. A great concept with intriguing potential. A sketch on a napkin, but seemingly viable. Like most ideas in this budding stage, it sat in the gap between “this could be something” and “I have no proof.” The thing that stands in that gap is research: who else is in this space, how large the market is, what it would cost to build, whether the idea can be defended once it’s visible. Done properly, that work takes a product manager weeks.
I am not a team. I am one person with an idea and a full set of other obligations. I could have spent a month of evenings assembling the picture myself, but with the work I’ve been doing in AI, I decided to run tests: I sat down on a Saturday afternoon and asked Anthropic’s Claude to help me do the research.
I typed out as many requirements as I could conceive in an hour. Hardly comprehensive, so Claude asked several follow-up questions and together, we arrived at a sufficient level of detail. Normally, I would expect more rigor but remember this was a test.
I did not expect what happened next. Another hour later, I had eight documents. Not bullet points, not a chat transcript I would have to translate into something usable — eight structured, properly formatted documents, each one the kind of artifact I would use to convince a leadership team that a product idea is viable and profitable.
- A Product Requirements Document (Markdown) — the concept specified in detail.
- A Technical Architecture document (Markdown) — how the product would actually be built.
- A System Diagram (HTML) — that architecture rendered visually.
- A Competitive Landscape report (DOCX) — who else occupies this space, and how serious each one is.
- A Market Analysis (DOCX) — the size of the opportunity, who the buyers are, and which way the trend is moving. This includes an impressive analysis of Total Addressable Market, Serviceable Available Market, and Serviceable Obtainable Market over my desired timeline.
- A Cost Analysis — what building it would actually cost, compared across different approaches like using a Scrum team or vibe coding.
- An IP Protection strategy — how to keep the concept defensible as it becomes visible.
- A Brand proposal — a screened short-list of brand names that harmonize with its purpose and market, and an analysis of brand availability across USPTO trademarks, available .com domains, the App Store, and Play Store.
Together those documents run to more than twenty thousand words. I want to be precise about the time, because the time is the entire point: this was an afternoon. A couple of hours. The same body of work, done the traditional way, takes weeks of a product manager’s life — or a five-figure invoice from an outside firm.
What struck me was not only the speed but the collaboration. I was not pressing a button and getting a generic template back. It was a deep conversation. Claude asked clarifying questions, pushed back when an assumption was thin, surfaced risks I had not thought of, and held the thread across all eight documents — so the cost analysis knew what the architecture had decided, and the IP strategy knew what the competitive landscape had found. Each document referenced the others. It behaved less like a tool and more like a colleague who had read everything and forgotten nothing.
There was a moment, somewhere in the middle of it, when I noticed I had stopped thinking of this as drafting and started thinking of it as deciding. The documents were appearing fast enough that the limiting factor was no longer how quickly I could write — it was how quickly I could make up my mind. A question would surface, I would answer it, and the answer would propagate into the requirements, the cost analysis, the market read. The work had inverted. The typing was free; the judgment was the scarce thing. That is exactly the right way around, and it is not how product work has ever felt to me before.
Here is the part that matters most to me. I am not going to describe the idea itself — that stays private for now. But I will say this: I sat down that afternoon with a hunch and stood up with a defense. These eight documents do not merely describe the concept; they argue for it. The market analysis showed the demand is real and growing. The competitive landscape showed the space is genuinely open. The cost analysis showed an affordable path to build it. The architecture showed that the hardest technical problem has a clean solution. Read as a set, they make a genuinely strong case that the idea is worth pursuing. It was no longer a hunch.
This doesn’t mean the work is finished, and it doesn’t mean I have handed off the thinking. The documents are drafts. I reviewed them all and they’re solid. The IP strategy will require review and input from a real attorney, and Claude was the first to say so. The IP document now opens with a disclaimer. The decision remains mine. The difference is that I now get to make it from a position of evidence.
I have thought a lot, since that afternoon, about what actually shifted. For most of my career the bottleneck in product work was never a shortage of ideas — it was the cost of investigating them. Investigation is expensive, so you ration it. You research only the ideas you are already fairly confident about, which means the surprising ideas, the ones that most need a fair hearing, rarely get one. Lowering the cost of investigation does not simply make the work faster. It changes which ideas get investigated at all.
I ran a second test a few days later. A different idea that’s been percolating in my mind for months, plucked from my Ideas backlog because I think it’s the most ambitious one with the greatest potential. This time, Claude pushed back much harder. The competitive landscape is fuller than I anticipated, so Claude laid out the facts for me. The idea still has potential, but now I know how steep the climb would be. At least now I have evidence about the cost to build and market, and the revenue potential.
The delta in PM practice is the thought I keep coming back to. A single afternoon used to buy me a vague sense of an idea. This time, an afternoon produced a decision-grade body of research. I learned that one key characteristic of the AI industrial revolution is that more ideas will get a fair shake at a fraction of the cost. In this case, I was ready to make the call in just hours. And I got there before dinner.