Assistant IA ECASTE

RAG Solution and AI Chatbot

Finding the right information in specifications, contracts and procedures stays slow, and mainstream AI tools send these documents to third parties.

Our mission

ECASTE has designed a RAG (retrieval-augmented generation) solution that indexes company documents in a local vector database and answers by citing the exact excerpt and source file. The language model is chosen per request from several LLMs, and the data stays stored in the customer's instance. The solution also lets you create a chatbot as a widget that embeds in a few lines on your own website, fed by documentation generated automatically from the site's content and then enriched manually.

Key points

• RAG on documents
• LLM choice per request
• Local data storage
• Sourced answers
• Embeddable chatbot widget
• Documentation generated from the website
• Manual enrichment
• Multi-project