Open source · DocuMind AI
Confluence answers, delivered in Slack
An open-source assistant that turns a Confluence space into a searchable knowledge base — ask the bot in Slack, get an answer sourced from your actual documentation.
- Hybrid retrieval: vector similarity over chunked, embedded Confluence pages, combined with keyword matching.
- Python 3.12 + MongoDB vector storage + DeepSeek API + a Slack bot; systemd configs included for production.
- Runs on a t3.micro with MongoDB’s free tier — roughly $5/month total, set up in about 10 minutes.
~$5/mo
Total hosting: EC2 t3.micro + MongoDB free tier
10 min
From clone to answering questions in Slack
OSS
Public repo — github.com/hengkysandy/documind-ai
The problem
Every team has the same loop: someone asks in Slack how to roll back a deployment, someone else digs through Confluence for the page that answers it, and next month it happens again. The documentation exists — finding it is the tax.
What it does
DocuMind ingests Confluence pages, chunks and embeds them into MongoDB, and answers questions in Slack with content pulled from the real docs — not a model’s guess. Retrieval is hybrid: vector similarity for meaning, keyword matching for precision.
Cheap by design
The stack was chosen so any team can afford to run it: one t3.micro, MongoDB free tier, DeepSeek for inference. About $5 a month, deployable in ten minutes, and open source so you can point it at your own wiki this afternoon.
🚀 I built and open-sourced DocuMind AI — because searching through endless documentation shouldn’t be a nightmare for your team. Transforms your Confluence docs into an intelligent Q&A system, with Slack integration — just @mention the bot with any question.
