Generative AI Experimentation

BNP Paribas (Innovation Lab)

3 months

AI Coordinator / Innovation Lead

Context

As part of BNP Paribas' innovation strategy, the team launched an experimentation program around generative AI. The objective was to explore potential use cases of AI (ChatGPT, Claude) to automate and improve internal processes, particularly in user support and document analysis.

Challenge

The banking sector imposes strict constraints in terms of data confidentiality and regulatory compliance. It was necessary to identify high value-added use cases while respecting security policies. The challenge was also to convince business teams of the potential of these technologies while managing fears related to automation.

Approach

I organized mob programming sessions with multidisciplinary teams (IT, business, compliance) to explore use cases. We prototyped several solutions: an internal chatbot, a regulatory document analysis tool, and a technical writing assistant. I wrote an innovation report detailing opportunities, risks and recommendations for scale deployment.

Results

3 POCs validated with business teams
Estimated 40% time savings on document analysis
Innovation report presented to COMEX with phase 2 budget approval
Creation of an internal community of 30+ employees trained in AI
Definition of an ethical charter for AI use

Learnings

Generative AI excels in synthesis and reformulation tasks
Prompt engineering is a key skill to develop in teams
Change management is as important as technology
IT/Business/Compliance collaboration from the start accelerates adoption