They said AI would kill SaaS boilerplates, but it's doing the opposite

They said AI Would Kill SaaS Boilerplates. It's Doing the Opposite.April 23, 2026 · 6 min readVince CangerDeveloper Relations @ Wasp. Creator of OpenSaaS.sh.When ChatGPT launched everyone was predicting that developers would soon be obsolete. This was about the same time we launched Open SaaS, our free, open-source SaaS starter. Using the same logic, you'd have expected that AI would have killed boilerplate starters by now. But we noticed the opposite trend. Open SaaS launched and grew during the vibe coding boom Open SaaS just crossed 14,000 GitHub stars and its entire growth happened right in the middle of the vibe coding boom. This surprised us. So we decided to talk to 40 of our users to learn why they turned to Open SaaS. A renewed Energy to Build.​ After conducting 40+ user interviews, we noticed a trend: AI was unlocking all kinds of new builders. Career devs, DevOps engineers, PMs, an ex-woodworker, a marketer with zero coding experience, and more were getting inspired to start on their Ideas. These were people who always had a SaaS idea but never had the time or skills, or people who were getting inspired to build new apps on top of AI. Regardless of which type they were, it was AI that unlocked them. "Two years ago I didn't believe I could build an application... So it's like kind of magic for me." — Leo, ex-marketer who built Messync Messync — built by Leo, an ex-marketer with zero coding experience "If there was no Wasp, I don't think I would even start building this." — Sergio, 20-year backend veteran who built CTOBox Another one of our users with no React or Node.js experience built a SaaS app and sold it to a major accounting firm for ~$100k. And the most surprising fact of all: about half of those interviewed were people who had never built a full-stack app before. AI gave them the confidence and inspiration to start, but a lot of them turned to Open SaaS to help them focus their efforts. AI Can Do 90%. The Last 10% Is the Hardest.​ Using Claude Design to design a landing page for Open Vibe, our new vibe code course AI tools are getting insanely good. With Claude Design you can scaffold a professional-looking landing page in about an hour. Slap it into Claude Code, add on more components and CRUD logic, and you're almost done. But it's things like Stripe webhooks, auth edge cases, environment management, deployment, and background jobs that really trip up builders. These aren't necessarily code problems, they're architecture problems. And AI is great at generating code within a clean, working architecture, but you've got to have one first. "Open SaaS made me feel secure, like I am not cutting any corners. Just using AI would make it harder to sleep at night." — Robbie, musician who built PeakMastering PeakMastering — built by Robbie, a musician. Trusted by YouTube, Spotify, Apple, and Fender. "With Next.js App Router, I was constantly fighting the LLM to get the syntax right. Svelte runes were just the Wild West." — TK Garrett, ex-woodworker who built PlotTree PlotTree — built by TK Garrett, an ex-woodworker turned developer Good Code Makes AI Code Better​ After talking with Open SaaS users, we heard the same thing over and over again: having a solid, clean codebase to start with drastically improved the perfomance of AI-coding agents and tools. And that makes sense, because when you use an opinionated, boilerplate codebase like Open SaaS, you've effectively already decided the architecture of your app for the AI and it…

This excerpt is published under fair use for community discussion. Read the full article at Wasp.