John & Tyler: Is Your Startup Idea Actually Good? AI, Vibe Coding, Steve Blank & Idea Validation

No guest this week — John and Tyler take on the question every founder is asking in 2026: how do you know if your startup idea is actually good, now that AI can build almost anything in a weekend? They start with Steve Blank's Lean LaunchPad 2026 write-up — his Stanford teams talked to 978 customers, and his verdict is that AI is making builders lazier about lean startup right when its principles matter most. From there: the "vibe-coding psychosis" (a slick MVP is not validation), what H1 2026's record $510B in venture funding really means when the top AI companies swallow most of it — including a single $110B OpenAI round — and why the PC, internet, and AI revolutions are the same movie of winners and losers. John lays out his test for a fundable idea: find a must-have workflow and drop its cost 90% or double its output. Plus: narrowing to a wedge (be the AI for HVAC installers, not "AI for sales"), why an orthodontist will never vibe-code his own practice software, using your first five customers as a laboratory, the Ash Maurya debate about letting founders run, and the discipline that saves you from burning two-thirds of your raise before the pivot. Also: an icebreaker on what AI actually did for them this week — car-hunting cron jobs, inbox zero, and the pickleball app that made John a local legend.(00:00:00) Cold open: AI is making founders lazy about lean startup(00:01:02) Welcome back — episode 58, no guest today(00:02:07) Icebreaker: What Did AI Do For You This Week?(00:10:08) Startup news: $510B in VC — the biggest half-year ever(00:14:26) Today's topic: what makes a startup idea actually good?(00:14:40) Steve Blank's Lean LaunchPad 2026: 978 customer interviews at Stanford(00:17:48) The vibe-coding psychosis: an MVP is not validation(00:22:40) The core question: is your idea a business opportunity?(00:29:42) Workflow economics: drop the cost 90% or double the revenue(00:33:46) Narrow your customer: the wedge, not the world(00:38:33) Horizontal vs. vertical: don't get crushed by the next LLM release(00:39:26) The orthodontist rule: nobody vibe-codes their core software(00:42:22) Use AI to speed up validation, not just building(00:43:04) "You still need to earn the right to build"(00:45:29) Five customers as your laboratory(00:49:47) The Ash Maurya conversation: let founders run?(00:55:19) The recipe: what makes a good idea in the AI era(00:56:23) Wrap-up: let us know in the commentsLearn more about us:https://startupignition.com/linksApply to be a guest on the pod:https://startup-ignition-podcast.paperform.co/Follow John & Tyler:https://x.com/trich_https://www.linkedin.com/in/johnrichards/Subscribe on YouTube:https://www.youtube.com/@StartupIgnitionFollow on X:https://x.com/startupignitionFollow on Instagram:https://www.instagram.com/startupignition/Follow on TikTok:https://www.tiktok.com/@startupignitionFollow on LinkedIn:https://www.linkedin.com/school/6458758/admin/dashboard/Follow on Facebook:https://www.facebook.com/startupignition/#StartupIgnition#LeanStartup#SteveBlank#VibeCoding#AI