Generative AI and ChatGPT: Combining Your Own Data With a Model

An introductory session from Afro Tech Paris on retrieval-augmented generation, explained without assuming any programming background.

The core question is practical: how do you point a language model at your own documents and get answers you can trust, rather than confident invention. The talk covers how retrieval changes what the model sees, and where the technique holds up against hallucination.

Applications covered include document search across a private corpus, HR and legal assistants, and internal knowledge bases. The session includes a live demonstration that turns a set of documents into a functioning assistant in front of the room.

The published agenda runs welcome and networking, then the talk, then the live demo, then an extended question and discussion block. Note that the agenda times printed in the body start later than the header start time, so the substantive content begins after the networking window.

No programming experience is required. Attendees are asked to bring a notebook or laptop. The session is hosted by Koffi Gilchrist Tossou. The exact address is inside Paris but is sent only to registered attendees.