Launching today

A self-hosted RAG chatbot boilerplate

Launching today

Own your RAG and LLM stack. GDPR compliance built-in.

10 followers

Own your RAG and LLM stack. GDPR compliance built-in.

10 followers

EuroRAG is source code for building AI chatbots over your documents. Upload PDF, DOCX, or spreadsheets. Ask questions. Get answers with source citations. GDPR compliance is architectural. Deletion cascades across database, vector store, search index, and disk. Per-provider data residency tracking. Strict EU Mode blocks non-compliant providers before any data leaves your server. Audit logging of every operation. Consent management with withdrawal support. Data export for subject access requests.

A self-hosted RAG chatbot boilerplate gallery image

Payment Required

Launch tags:Software Engineering•Developer Tools•Artificial Intelligence

Launch Team

Tines Build agents & automations integrated across your workspace

Promoted

Maker

📌

Hey PH! I'm Dominykas, the developer behind EuroRAG — a self-hosted RAG chatbot you buy as source code and deploy on your own EU infrastructure. The problem: no RAG framework out there provides architecture and tooling for GDPR compliance. "Just deploy it in the EU" isn't compliance, you need deletion cascades that actually erase data from every storage layer, audit trails, consent management, and data residency controls that block non-EU providers at the code level. EuroRAG builds all of that into the architecture: • Delete a user → cascades through the database, ChromaDB vectors, the BM25 search index, and files on disk. Soft-delete makes it instantly invisible; a scheduled purge makes it physically gone. • Strict EU Mode (on by default) → blocks OpenAI, Groq, and other US-based providers before any data leaves your server. • Audit logging on every consent change, data export, deletion, and admin operation — the event is logged, never the content. • Hybrid search → BM25 keyword + vector similarity, fused with Reciprocal Rank Fusion. One shared knowledge base for the whole team by default, strict per-user isolation as a config switch. Stack: FastAPI, SQLAlchemy (async), HTMX, ChromaDB, Ollama or any EU-hosted API (Mistral, Scaleway). No React, no Node.js, no webpack, if you know Python, you can read and modify every part. Why paid source instead of open source: the GDPR machinery, admin tooling, test suite, and docs are the product, and a perpetual license (€179 Professional / €89 Starter) is how one developer sustains maintaining it. You buy once, get every line, modify anything, ship it to your clients. The code keeps running whether or not you ever pay again. Feel free to ask me anything.

Report

1d ago

honestly the cascade deletion story is really well thought out, one thing that would make this even better is a built in redaction layer for PII before chunks get embedded, basically detect emails phone numbers and IBANs in the source documents and either mask them or replace them with placeholders so nothing sensitive ever lands in the vector store

Report

2h ago

Finally tried a RAG setup that takes GDPR seriously. The cascade delete across Postgres, the vector store, and the search index actually works as advertised, which is rarer than it should be. Strict EU Mode is a nice touch for anyone who has dealt with compliance audits.

Report

1h ago

Cascade deletion across all the stores at once is genuinely handy, not having to script cleanup for each layer separately. EU Mode blocking non-compliant providers before anything ships out is a nice touch, honestly saved me from a footgun I'd probably have missed.

Report

27m ago