Hi HN! I'm Aryan, the maker of TemplateAI. I tinkered with a bunch of AI side projects in 2023 and realized I'm repeating a lot of the same stuff: calling text and image gen models, setting up vector search and generating embeddings, storing everything, building components for chats and generated content. Plus dozens of hours spent building the rest of stack – auth, payments, dashboard and landing etc.

I packaged it all into one NextJS template so others can skip past the boilerplate and go straight to the AI features.

The template is designed for three big use cases: image generation, text generation, and vector search / retrieval. It uses a slightly opinionated stack: TypeScript-first with NextJS (App router), Supabase as the database layer, OpenAI's LLMs for text generation, Replicate API for image generation, and tools like LangChain and Vercel ai sdk to simplify RAG and response streaming.

For authentication and vector databases, I stuck with Supabase Auth and pgvector to avoid adding yet another piece to an already-complicated stack. (There's certainly other vectorstores out there that would speed up retrieval that I'm considering adding to the template.) Also wanted to make sure everything's launch-friendly, so I added in landing page components, a Stripe integration, and more.

Please check out TemplateAI and share your thoughts and feedback, thanks!