A promising idea is still an assumption. Test the customer need, and reduce the uncertainty, before committing to a major build.
It is easy to picture the finished product and then look for people who might want it. The more useful order is the reverse. Treat the idea as a hypothesis: someone has a problem, they care enough to change how they work, and this particular approach would help.
You can test that without funding a large build. Talk with the people who live with the problem. Show a sketch or a simple mock-up and watch where they hesitate. Run a manual pilot, where a person does the job the product would later do. Only when those steps still look promising is a minimal usable version justified.
Decide in advance what would count as a useful signal. That might be adoption, repeat use, fewer delays or errors, or a willingness to pay. A polite “that sounds interesting” is not the same as evidence.
Then make a decision the evidence can support: continue, revise, narrow the audience, or stop. Stopping early is a successful outcome when it protects time and money that would otherwise go into the wrong thing.
Who would actually feel the difference if this idea worked—and have you heard it from them?