EP992: Preparing Banks and Insurers for the Agentic AI Era

The next frontier in banking and insurance isn't just adopting AI—it's executing it at scale with real-world impact. Alfred Gautier of Asensio argues that the biggest barrier to production-grade AI isn't technology, but legacy infrastructure and regulatory caution. He warns that without a solid foundation, even the most advanced AI systems will fail, likening it to building a mansion on a weak foundation. The real value of AI in BFSI lies not in flashy chatbots, but in embedded, mission-critical applications: slashing false positives in anti-money laundering by up to 86%, enabling one-minute insurance claim settlements, and automating end-of-month financial closes. Yet, Gautier stresses that agentic AI must be rolled out with extreme care—especially in life and medical insurance, where human empathy is irreplaceable. He introduces a 'mentor model' for AI: start with zero confidence, let it learn under supervision, and only grant autonomy when it consistently outperforms humans. The future isn't about replacing people, but empowering them with AI that’s trained on real data, certified for compliance, and built with guardrails—because in finance, one misstep can cost billions, not just a few dollars.