What we built: A multi-tenant no-code platform that helps businesses create custom ChatGPT Apps. When someone asks “find me a legal firm in Austin,” businesses using our platform appear with their own branded app - complete with UI components, interactive elements, and real-time information. Not random results from old training data, but actual custom experiences with each business’s branding.
Try it: We’re in limited beta with founding members. You can create and test an app yourself in ChatGPT’s Developer Mode - takes just a few minutes (email me for examples - don’t want to spam here).
Technical details: We generate MCP servers that expose business services as tool contracts. When ChatGPT’s function-calling matches user intent, we return real-time data through OAuth-authenticated sessions (for now we’re not releasing OAuth to our users so they can test with less friction).
What was hard: Building a no-code abstraction layer over MCP servers, OAuth flows, and JSON schemas while maintaining flexibility for custom experiences. Every business wants different functionality - appointment booking, product catalogs, lead forms - but ChatGPT’s Apps SDK has strict requirements. We had to create a system that generates compliant tool contracts and web components automatically while still allowing businesses to customize their UI and branding.
What surprised me: Most businesses / startups have no idea ChatGPT Apps even exist. The October 6 launch flew completely under the radar outside tech circles.
What I’m unsure about: Are we too early? Will businesses pay for AI discovery before it’s proven?
We’re onboarding founding members now - businesses / startups that want to establish AI discovery presence before their competitors.
If you run a company or know someone who should be discoverable when people ask AI for recommendations, join our waitlist at [noodleseed.com].
Would love feedback on the approach and whether anyone else is tackling AI discovery differently.