NG Banking Innovation Summit · 2024
How AI agents transform banking operations
From five weeks to 20 minutes: a real complaints workflow, rebuilt with multi-agent AI
What does it actually take to use AI agents in real banking operations?
This session uses a complaints investigation and redressal workflow in wealth management as the worked example. The process previously took five weeks. With a multi-agent system, human-in-the-loop review, and governance controls built for production, it runs in 20 minutes.
The session covers:
- AI agents for banking and financial services operations
- single-agent vs. multi-agent systems
- supervisory, hierarchical, and custom agent structures
- complaint investigation automation in regulated environments
- unstructured data across emails, PDFs, reports, and internal systems
- citations, traces, and data lineage for explainability
- pilot-to-production lessons on accuracy, ROI, and scalability
- governance and control for enterprise AI in production
The point of the session is to show what it takes to get an AI agent out of a demo and into daily operations.