All case studies
RAG & knowledge systems

ShamelaGPT

A retrieval-augmented research assistant over one of the largest digital repositories of Islamic texts. Its centrepiece is fact-checking: submit a claim — including as an image — and the system cross-references it across the corpus, returns a verdict grounded in what it found, and links every source so the reader can verify it independently.

The challenge

Researchers and students in Islamic studies needed an intelligent way to search and query the Shamela digital library, one of the largest repositories of Islamic texts. Searching manually through thousands of scholarly works was slow, with limited ability to find relevant passages across different texts and languages.

What we built

We built a retrieval-augmented generation system over the Shamela digital library, with multilingual support and a conversational interface designed for scholarly research rather than general chat. Its central capability is source-grounded fact-checking.

  • Fact-checking from an image. Submit an image of a claim presented as a hadith or similar, and the system reads it and checks it rather than taking it at face value.
  • Cross-referencing across books. The claim is checked against multiple works in the corpus, not answered from a single source.
  • Verdict with citations. Returns a final verdict based on the documents it found, with links to each one so the reader can cross-verify independently.
  • Semantic search. Retrieval across thousands of Islamic texts by meaning rather than keyword match.
  • Multilingual support. Processing and querying in multiple languages.
  • Conversational interface. Natural-language queries for complex research questions.
  • Large-scale processing. Handling one of the largest digital libraries of Islamic content.

Results

Fact-checked
against the corpus, with a verdict and links to every source
Thousands
of Islamic scholarly works made semantically searchable
Live
in production at shamelagpt.com

Technology

  • RAG
  • Semantic search
  • Multilingual

Have a problem like this one?

Tell us what you are trying to automate, extract, analyse or build, and we can work out what an appropriate system would look like.

Start a conversation More case studies