2026-05-06 • Casey Alvarez • 13 min
Discovery Isn’t a Workshop Deck—It’s Defense Against Self-Lying Founders
Continuous discovery protects builders from seductive ideas backed only by caffeine and optimism. Here’s how signal hygiene beats brainstorming theater.
Three summers ago I incubated an AI workflow assuming procurement managers secretly hated spreadsheets more than they hated procurement theater. The hypothesis sounded cinematic until interviews politely surfaced reality: nobody loses sleep over CSV merges—they lose sleep about audits.
Automating discovery doesn’t replace instinct; it stops founders from narrativizing cherry-picked anecdotes into destiny arcs.
Treat signals like instrumentation with uptime graphs
Noise spikes whenever novelty floods timelines—GitHub trending repos, AI newsletters breathlessly claiming paradigm shifts. Separate novelty spikes from sustained irritation loops bubbling inside niche forums where practitioners swear casually.
- Blend qualitative shards (verbatim frustrations) with quantitative breadcrumbs (GitHub issue velocity, repeated Reddit threads)
- Document counter-evidence mercilessly—the memo defending why NOT to build matters more than cheerleader Slack screenshots
- Rotate sources deliberately so algorithmic drift doesn’t trap you inside identical bubble chambers weekly
Ranking isn't cruelty—it's compassion toward engineers whose calendars deserve protecting.
Scorecards translate vibes into debate-able bets
Evidence density beats vibes-only prioritization. Differentiation cannot merely sound poetic—it needs comparative anchors (“why won’t incumbent X ship this quarter?”). Feasibility weighs skeleton-team realities versus fantasy infinite-contractor timelines.
Auto-enqueue top-ranked candidates only after humans sanity-check scoring skew—otherwise automation confidently executes brilliantly optimized nonsense.
Discovery ceremonies decay unless rituals repeat boringly
Weekly reruns beat quarterly ceremonies because markets mutate faster than quarterly OKR grids refresh. Treat stale rankings like flaky CI—surface amber warnings instead of silently pretending priorities froze correctly.
Serious teams ritualize retiring hypotheses loudly—celebrate killing ideas compassionately so teammates pitch sharper replacements without fearing reputational tombstones.