RefChat features

A research assistant built for scientific rigour and data sovereignty.

Why RefChat?

Six pillars that set it apart from a generic RAG or a ChatGPT-dressed-as-PDF-search.

🔒

Absolute confidentiality

Your documents never leave your machine in local mode. Cloud mode available if you'd rather offload the heavy lifting to our European servers.

🎯

Precise hybrid search

Multilingual semantic search (E5-large), BM25 keyword, and cross-encoder reranking. More accurate than a basic RAG, with sources cited explicitly.

🧠

5 smart modes

RefChat automatically detects what you want: question, summary, references list, author search, fact-check.

📄

Advanced OCR & parsing

Scanned PDFs? Old image-only reports? EasyOCR + GROBID extract text, metadata and scientific structure.

🗂️

Automatic topics

BERTopic clustering organises your library into coherent themes. Rename, merge or split them to fit your mental model.

🌍

Multilingual by design

Ask a question in English about a French article (or the other way around) — RefChat finds the relevant content regardless of language.

5 query modes detected automatically

You write your question in natural language — RefChat infers what you actually want.

ModeTriggerBehaviour
Question(default)Narrative answer with sourced citations
Summary"summarise", "synthesis"Synthesis of one or several articles
References"which publications"List of relevant articles with explanations
Author"articles by", "works by"Author search via OpenAlex
Fact-check"check", "is it true that"Verifies claims against your bibliography

RefChat in pictures

A glimpse at the interface — from sourced chat to thematic exploration of your library.

💬 Screenshot: chat with clickable citations

Sourced conversation

Every answer cites the exact passages of the underlying articles, with direct links to the PDF.

🗂️ Screenshot: BERTopic theme map

Thematic mapping

Visualise the main axes of your library, automatically grouped by topic.

⚙️ Screenshot: indexing in progress (GROBID + OCR)

Transparent indexing

Detailed progress of GROBID parsing, OCR, embeddings and topic modelling.

Who is it for?

Built for demanding environments where confidentiality and rigour come first.

🔬 R&D and academic research

State-of-the-art synthesis, cross-referencing hundreds of articles. Accelerate your literature reviews.

📊 Technology watch

Sector reports, patents, publications. Ask strategic questions with zero risk of leak.

📚 Industrial knowledge mining

Surface the information buried in decades of internal reports (OCR + RAG).

🏛️ Consultancies and expertise

Ideal for organisations handling sensitive documents (legal, medical, geoscience).

Try it on your own library

30-minute live demo, or a trial run on a sample of your PDFs, under NDA if needed.

Request a demo