Prototype with AI before committing to a roadmap.

Build a throwaway version in days to validate the idea before quarter-long builds.

Why this matters

The most expensive bug is shipping the wrong feature. AI coding tools have made it cheap to build a rough working prototype in a day or two, enough to put in front of real users and see what they actually do. That feedback loop saves months later.

What to do

  • Before spec'ing a multi-quarter project, scope a 2–3 day prototype.
  • Use Claude Code or v0 to build a working version with mock data.
  • Get it in front of 3–5 real users.
  • Decide whether to invest based on what they did, not what they said.
  • Then throw the prototype away; production code has different requirements.

Common pitfalls

  • Don't let the prototype become the production code; shortcuts compound.
  • Don't show the prototype to stakeholders without context; they'll mistake it for the real thing.
  • Don't skip user testing because the prototype "feels right" internally.
Start a Project or email us

Want help applying this to your setup? Tell us where you're stuck.