I've been re-reading bits of Ryan Singer's book Shape Up about how they develop products at basecamp. This isn't the first time I've been back to it, and each time a different bit stands out.
This time, it's the concept of getting roughly shaped ideas to a betting table – figuring out the problem you're trying to solve, through discovery-focused research, then sketching deliberately rough, high level ideas for solutions, for discussion with a small group. Shaping ideas for things to build.
Keeping ideas deliberately rough is important. Ryan recommends using fat-markers and avoiding details like actual copy or graphic elements.
As well as the book above, I'd recommend this flowchart as a quick distillation of the process.