Kensho

Most product tools help a team move faster once it knows what it wants to build. Kensho is built for the earlier and harder question: what else might be possible, and which possibilities have earned deeper commitment?

A transformer language model can generate a plausible product concept in seconds. Ask for ten and it will produce ten. What it does not produce is a map: which regions of possibility have been explored, how the candidates meaningfully differ, which tradeoffs recur across them, or where the search should move next. A longer list may provide more options, but it does not necessarily produce better discovery.

Kensho is a product-discovery workbench for turning a problem or opportunity into an ongoing program of exploration. Instead of carrying one favored concept through a familiar funnel, it helps construct, compare, and evolve many possible products while preserving the differences among them. The goal is not to return a single machine-selected winner, but to create a field of possibilities that people can investigate and learn from before deciding where to commit.

An exploration begins with the opportunity itself: who experiences it, what conditions shape it, which constraints cannot be ignored, and what meaningful variation might look like. Kensho then helps widen the field without dissolving it into an unstructured flood of ideas. Candidates can be compared, related, revisited, and used as starting points for further exploration. As the field develops, the important object is no longer any individual idea but the evolving map: promising regions, recurring patterns, unexplored gaps, surprising outliers, and the paths by which each possibility emerged.