08-14-2026, 01:58 PM
The fastest path I've found, before writing a single line of code:
1. Talk to 10 real potential users – not friends. Ask about their current pain, not your idea. If 3+ describe the same burning problem, you're onto something.
2. Pre-sell with a landing page – one page + a "Join the waitlist" button. Real signups = real demand.
3. Run a concierge MVP – deliver manually by hand what the product would automate, for 1–3 beta users. If they keep paying, automate.
4. Set a kill criterion – e.g. "fewer than 20 waitlist signups in 3 weeks → pivot."
Most AI side projects die from building first and validating never.
What's the cheapest validation experiment you've run?
1. Talk to 10 real potential users – not friends. Ask about their current pain, not your idea. If 3+ describe the same burning problem, you're onto something.
2. Pre-sell with a landing page – one page + a "Join the waitlist" button. Real signups = real demand.
3. Run a concierge MVP – deliver manually by hand what the product would automate, for 1–3 beta users. If they keep paying, automate.
4. Set a kill criterion – e.g. "fewer than 20 waitlist signups in 3 weeks → pivot."
Most AI side projects die from building first and validating never.
What's the cheapest validation experiment you've run?
👍 Like0❤️ Love0😂 Haha0🔥 Fire0👏 Clap0⚡ 0