The problem
I invest for the long term, and I kept running into the same wall. To judge one company I had to read reports, compare ratios, check the price history and weigh a dozen opinions. After hours of it I was often still unsure.
The tools I found sat at two extremes. Some showed everything and explained nothing. Others gave a confident answer with no way to see where it came from. I wanted something in between: a clear answer, with its reasons in view.
Who it is for
People who choose their own stocks and hold them for years. They are willing to do research, but they are not analysts, and many are near the start. That last point shaped a lot of later decisions.
Defining the core
I reduced the product to one action and three questions.
The decision that shaped the product
The decision
The rating comes from a rules-based model. AI explains the result and does not decide it.
Early on, the quick route was to let an AI model read about a company and write the whole analysis. I wanted the rating to be deterministic and transparent. The same inputs had to give the same result, and a rating had to be traceable to rules that can be tested.
That choice cost flexibility and took more work. It is also why the product can score about a thousand companies every night, and why a change to the model can be checked exactly before it is released.
What I left for later
Deciding what not to do mattered as much as building.
- A short menu. The main navigation shows the few things people come for. Everything else sits under "More".
- No feature for its own sake. When the list of ideas grew, I set two priorities, bringing the right visitors and letting them see value fast, and paused what did not serve them.
- No shortcuts on reach. Broad advertising and mass-produced pages were put aside in favour of channels where I could learn from the people arriving.
Launching in stages
I widened the audience one step at a time, so each larger group arrived at a product that a smaller group had already used.
After launch
Real users changed my priorities. Two parts of the product got focused work of their own, and each has its own case study:
The result
- A live product with over 100 registered users and its first paying customers, built independently.
- About a thousand US stocks scored automatically every night.
- Paid plans and a free trial, live since spring 2026.
- One scoring core that the product's tools are built on.
What I took from it
- Reduce the product to one action before adding anything around it.
- Grow the audience in steps, and learn from each one.
- What I leave out is a product decision too.
