Two modes, your choice
The hybrid search engine and the interface are identical in both modes. What changes is where your index lives, and where the LLMs run.
🔒 100% Local mode
- Absolute confidentiality: no data leaves your machine
- No external server involved in processing
- Local LLM via llama.cpp (Mistral, Llama GGUF models, etc.)
- Local indexing (embeddings + BM25 + reranking)
- Ideal for sensitive data (industry, legal, medical)
- Less powerful PC: indexing and queries are slower
- Recommended: 16 GB RAM minimum, NVIDIA GPU for good performance
☁️ 100% Cloud mode
- Online LLM (Mistral API): top-tier answer quality
- Indexing offloaded to our OVHcloud servers (France)
- No hardware constraint for the user
- Greatly accelerated indexing and query speed
- Data encrypted in transit (TLS), hosted in the EU
- Ideal for large volumes, light setups, distributed teams
Three steps, that's it
From your PDF folder to your personal research assistant, in minutes.
Point at your library
Tell RefChat where your Zotero, Mendeley or PDF folder lives.
Smart indexing
Multi-threaded pipeline: GROBID parsing, OCR, semantic chunking, embeddings, BM25, topic modelling.
Chat with your articles
Ask your questions in natural language. Answers sourced with clickable citations.
How to choose?
A few practical rules of thumb depending on your profile and constraints.
| You are… | Recommended mode |
|---|---|
| A researcher with a recent PC + NVIDIA GPU | Local — solid performance, no recurring cost beyond the licence |
| A consultancy handling confidential matters (legal, medical) | Local — mandatory, no data should ever transit |
| An industrial R&D team with patents or internal reports | Local or Team (hybrid) depending on sensitivity |
| A PhD student with a modest PC and 5,000+ articles to index | Cloud — otherwise local indexing could take days |
| A library of 30,000+ articles | Cloud — local becomes uncomfortable past ~20,000 |
| A distributed team sharing an index | Cloud or Team — centralised index on OVHcloud FR |
The right reflex
Start in local mode on a sample (a few hundred PDFs) to validate the use case. Switch to cloud or hybrid (Team plan) if your volumes or hardware constraints justify it — no manual re-indexing, RefChat handles the migration.
Wondering which mode fits your case?
30 minutes with us is enough to point you in the right direction — no strings attached.
Get in touch