What happens when a bank decides that AI and IP are so strategic they must be built in-house - then actually follows through for more than a decade?

In this episode of The Data Playbook, Dr. Barak Chizi, Chief Data & Analytics Officer at KBC Group, joins Kris Peeters to reveal how KBC built one of Europe’s most mature AI organisations and what it took to bring Kate, their AI assistant, to life, and keep her evolving for 5 years.

You’ll hear how KBC:

• Grew from early machine learning to 2,000+ AI use cases in production

• Developed an AI-driven anti-money laundering platform and commercialised it for other banks

• Scaled Kate, now celebrating 5 years and upgraded with GPT.

• Uses the U-model to govern AI safely from idea to production

• Keeps ROI at the centre of every AI project

• Stays vendor-independent while still leveraging hyperscaler LLMs

• Builds diverse, high-calibre AI teams with a rigorous recruitment approach

• Explores soft logic and modelling customer intent as the next frontier of financial AI

If you want to understand how to turn AI from experiments into a true competitive advantage, this conversation is your playbook.

🌐 More at www.dataminded.com and subscribe to our channel.

Show notes:

The Foundation of Soft Logic👉 https://link.springer.com/book/10.1007/978-3-031-58233-2

Dan Ariely – Predictably Irrational👉 https://www.amazon.com/Predictably-Irrational-Revised-Expanded-Decisions/dp/0061353248/

⏱️ Chapters

00:00 – Intro to The Data Playbook & today’s guest01:15 – Barak’s backstory: 25 years in AI & high-dimensional data03:02 – What a CDAO does at KBC & enabling 24/7 AI-assisted service04:55 – Towards continuous, machine-supported customer journeys06:37 – The U-Model: KBC’s framework for data & AI projects08:35 – Flagship AI products, finite project lifecycle & retraining10:07 – Prioritising AI use cases across 5 countries12:31 – ROI mindset, conservative risk culture & data as an asset14:21 – Why KBC keeps AI in-house & limits external consultants18:17 – Beyond data warehouses: from reporting to prediction22:21 – AI-driven AML platform & the creation of SKY25:30 – Patents, AI IP and KBC’s competitive positioning27:25 – Generative AI at KBC since 2018 & early transformer experiments29:11 – Pragmatic tech choices: LLMs vs ML vs simple automation31:42 – Avoiding GenAI hype and focusing on customer value33:03 – Why KBC built Kate: 24/7 banking & impatient customers35:28 – From FAQ bot to execution engine: Kate’s end-to-end capabilities37:07 – Customer reactions, branches vs digital & Kate’s 2026 roadmap39:24 – Multi-LLM strategy, vendor independence & design partnerships40:44 – Inside Kate’s architecture: NLU, open source & KBC-built layers42:37 – Proactive AI: timing, context and personalised offers44:51 – Soft logic, consciousness & modelling customer intent49:19 – Building a diverse, 24-nationality AI team at KBC51:37 – Recruitment process, tests & how candidates are evaluated55:21 – What KBC looks for in modern data scientists57:15 – Lessons after 10 years at KBC & book recommendation