Chatbots that read your docs

AskMyBot lets anyone build a chatbot that answers from their own material instead of the open internet. You sign in, create a named bot, upload your documents — a policy handbook, product docs, lecture notes — and from then on the bot answers questions using what's in those files, and points back to where it found the answer. Every account is isolated: your bots and your knowledge are yours alone.

Answering is a retrieval-augmented-generation (RAG) loop. When you ask something, the question is turned into a vector and matched against your documents in a vector index; the closest passages are pulled back as context, combined with the recent conversation, and handed to the model, which writes the reply. The answer streams back word by word, so you watch it form in real time rather than waiting for the whole thing.

Under the hood it's a deliberately serverless, two-part system. A Next.js app on Vercel handles the interface, login and uploads; a separate Cloudflare Worker does the heavy retrieval and generation close to the data. Documents live in Cloudflare R2, their text and your bots' prompts in Cloudflare D1, the searchable vectors in Cloudflare Vectorize, and your account in Neon Postgres — so each concern scales on its own and the whole thing runs without a server to babysit. It is deployed and live at askmybot.me.