User research "in the wild" can lead to surprises and learnings, especially critical for early-stage startup founders who need to test and reject or double-down on hypotheses to build the right product for the right market. This practice, refined over a decade at startups like Code Yam and Kamcord, draws from the Design Sprint process by GV and the GV Research Sprint by Michael Margolis. The author uses variations of these methods to get real feedback from potential users quickly, testing risks, assumptions, and value propositions in low-risk ways.

Unlike corporate design sprints that often yield no real results, this approach focuses on actionable insights from small, representative groups. New AI tools can accelerate hypothesis sharpening and testing, but the author stresses they should not replace direct human conversations—especially since even AI-agent products involve human buyers or operators. Key use cases include testing new product ideas, value propositions, landing pages, user group understanding, problem validation, willingness to try or pay, and UI/UX effectiveness.

A counter-intuitive insight is that user research helps founders realize when they’ve failed to meet objectives, such as when an idea doesn’t resonate or the market is too small or inaccessible. The core principle remains: founders must be "in the arena," talking to real people who might become users or buyers.