nikhil61191

We just open-sourced Unstruct, which we describe as a “Supabase for AI.” It’s a Django/PostgreSQL backend that handles the common needs of AI applications:

- Data connectors for file uploads (S3, Google Drive, etc. on the roadmap) - LLM integrations (OpenAI by default, extensible for others) - Vector DB support for embeddings and semantic search - A REST API interface so your front-end or other services can interact easily Why we built it: After talking with multiple AI founders, we saw everyone reinventing the same backend plumbing—connecting data sources, hooking up vector databases, orchestrating LLMs, and providing some sort of REST layer. Unstruct aims to save teams from that repetitive setup so they can focus on their app’s unique logic.

It’s released under MIT. Check out the GitHub repo for docs, a quickstart, and to see how it works. We’d love feedback or contributions—thanks for taking a look!