Context
Content has to land in two places now — Google rankings, and AI answers such as ChatGPT, which the product addressed as GEO. That work is otherwise done by hand, over and over.
AI SaaS — 2026
VibeSEO
SEO content is a manual loop: research, write, publish, repeat. VibeSEO automated that loop end to end. I co-founded it and owned product and engineering as CTO.
What I built
A user signed in with Google, connected a site, and got a keyword roadmap. From that roadmap the platform generated SEO/GEO articles and auto-published them to the connected CMS. In front of the product sat a free GEO audit tool that checked a site’s crawler access, schema, technical SEO and llms.txt with no login.
Architecture
The stack was NestJS and TypeScript on the backend, Prisma for the data layer, Redis, and a multi-LLM layer for generation. Publishing depended on a system the product did not own: an article was only done once the target CMS had accepted it.
Challenges
Two things had to be true for this to work. First, unattended generation: articles were generated and published with no person in the loop, so every step had to be validated and repeatable. Second, no single model vendor could be a hard dependency; generation ran through a multi-LLM layer.
Outcome
The platform ran in production until July 2026, when the project wrapped up.
Results
Built the product and the teamIn production until Jul 2026Wrapped up · Jul 2026
Discuss a project Have a product to ship?
Send the idea, the repo or the half-built thing. I answer with questions first, then a written scope. Hiring? Tell me which role.