Belfius Insurance’s Head of Data & AI, Hannes Heylen shares how his team scaled GenAI - from a fraud detection flywheel to “Nestor,” a claims copilot that speeds summaries, completeness and coverage checks. We unpack AI agents in the claims flow, build-vs-buy decisions, and why content/data governance drives LLM quality. Plus: a pragmatic delivery mantra - make it work, then right, then cheap - for CIOs, CDOs and Heads of Data.

What you’ll learn

• How to pick first AI cases that prove €ROI (fraud models)
• Designing a claims copilot: summarization, completeness & coverage checks
• Where AI agents fit (GenAI + ML + humans) across the claims flow
• Build vs. buy in 2025: foundation models, vendor flexibility, cost control
• Content/data governance as the make-or-break for LLM apps
• “First make it work, then right, then cheap”: an AI operating model for CIO/CDO

Guest: Hannes Heylen, Head of Data & AI, Belfius Insurance

🌐 More at ⁠www.dataminded.com⁠

Chapters:

00:00 Why AI now in financial services

06:30 GenAI’s impact on text-heavy insurance processes

18:40 AI agents across claims

31:00 Governance > model tweaks

38:00 Fraud detection: the € case

41:30 Claims copilot (“Nestor”) & lab-to-prod

55:00 Lessons for CIOs/CDOs

Topics: ROI-first use cases • Claims automation • AI agents (GenAI + ML + human-in-the-loop) • Governance • Vendor flexibility & costs

In this episode of The Data Playbook, we go inside imec, one of the world’s leading semiconductor research institutes, to explore how they scale data governance, self-service, and innovation in one of the most data-intensive environments on Earth.

Our guest, Wim Vancuyck, Manager of ICT for Data & Research Enablement, leads imec’s data strategy - bridging IT, researchers, and business to accelerate R&D through digital solutions. Wim’s mission: make imec a data-driven research organisation that turns raw measurements into insights and intellectual property faster and more securely.

Wim explains how imec built a research data platform that empowers thousands of scientists through:

• Purpose-based access control, linking people, platforms, and data assets
• Four self-service workbenches for Power BI, Data Engineering, Data Science & AI, and Application Development
• A clear platform vision built on efficiency, scalability, and reliability
• A governance model that supports both compliance and creativity
• A pragmatic stance on shadow IT: embrace, standardise, and professionalize it
• A bottom-up adoption strategy driven by early adopters and community engagement

He also discusses his evolution from technical architect to data leader, and what it takes to manage change in a 5,000-person R&D organisation, balancing technical depth with people leadership.

🎙️ Guest: Wim Vancuyck - Manager ICT, Data & Research Enablement, imec

🌐 More at: www.dataminded.com

#DataPlaybook #IMEC #SemiconductorR&D #DataGovernance #PurposeBasedAccessControl #DataMesh #PlatformEngineering #SelfServiceAnalytics #CIO #CDO #DataLeadership #ResearchDataPlatform #DataStrategy

How do you turn governance from a bottleneck into a business accelerator - in a bank?

In this episode of The Data Playbook, host Kris Peeters talks with Jan Mark Pleijsant, Senior Data Strategy & Governance Advisor at ABN AMRO Bank, about their move from 300+ dispersed data owners to 15 clear, business-aligned data domains and what it took to make federated governance work in a highly regulated environment.

You’ll learn:

• Why “governance police” fails and how a federated model balances autonomy with oversight.
• How ABN AMRO defined domains by the nature of data (customer, loans, payments) rather than shifting business units driving stability across reorganisations.
• What “governance by design” looks like in practice (policies, standards, data lineage, quality, clear ownership).
• The evolution from golden (source-aligned) datasets to consumer-ready data products (“pre-packaged salads”) that speed time-to-insight.
• Why executive sponsorship, domain maturity, and delivery discipline are the real success factors.
• How to measure domain maturity, prioritize critical reports (e.g., regulatory & exec dashboards), and avoid endless debates over definitions by making differences explicit.

Who should listen: CIOs, CDOs, Heads of Data, and Data Leaders building scalable, compliant data platforms in complex organizations - especially in financial services.

Chapters:

00:00 Intro & context (POA Summit)

03:12 Why banks need strong data governance

07:45 Federation vs. centralization (and pitfalls)

13:10 From 300 owners to 15 domains

19:40 Designing domains that don’t break with reorgs

26:05 Data products vs. datasets—what changed

33:20 Governance by design: policy → product

40:05 Measuring maturity & building momentum

47:30 Risks, success factors, and the next 24 months

🌐 More at https://www.dataminded.com/

#DataGovernance #FederatedGovernance #DataDomains #DataProducts #DataLeadership #BankingData #ABNAMRO #DataStrategy

What does it take to scale data products across an organization? In this episode of The Data Playbook, we sit down with Simon Harrer, CEO and Co-Founder of Entropy Data, recorded live at the POA Summit 2025 in Stuttgart.

Simon unpacks his journey from developer to founder, the creation of Data Mesh Manager, and why data contracts are becoming the backbone of modern data governance. We dive deep into:

• The evolution from consulting to product-based data companies

• How data products and contracts drive interoperability

• Why AI and MCPs will redefine how data is shared and governed

• The future of Data Mesh and the rise of data marketplaces

A conversation packed with real-world lessons for Data Leaders, CIOs, and CDOs driving digital transformation.

🌐 More at www.dataminded.com
🔗 Resources from the episode:

• Entropy Data

• Data Mesh Manager

• Data Contract CLI

• Data Product MCP

• Entropy Spin-off Story

In this episode of the Data Playbook podcast, we dive into how publiq, the organization behind Belgium’s largest cultural event database, is building RADAR, an AI-powered framework that enriches and structures event data at scale.

Host Kris Peeters is joined by Sven Houtmeyers (CTO) and Elia Van Wolputte (Data Scientist) from Publiq, who share how their team uses LLMs, semantic parsing, and linked data to improve search, recommendations, and user experience, all while respecting publiq values like privacy, transparency, and digital inclusion.

Topics covered:

• Why unstructured data is a challenge for cultural event discovery
• How RADAR leverages LLMs for smarter enrichment and entity resolution
• From batch jobs to live services: Scaling across GCP and AWS
• Designing ethical AI in the public sector: avoiding filter bubbles and over-personalization

This is a behind-the-scenes look at how public organizations can use modern AI tools, not to manipulate users, but to empower them.

🌐 More at ⁠www.dataminded.com

🎙️ In this episode, host Kris Peeters talks with Jelle De Vleminck, consultant at Dataminded, about what it really takes to build a data platform that people actually want to use.

Together, they explore:

• What separates a platform from just “a bunch of tools”
• How to reduce cognitive load for developers
• The 5 biggest mistakes in platform design
• Why adoption matters more than features
• How data contracts, products, and cloud IDEs improve usability
• Why enabling your users beats controlling them

If you’re building internal tooling or scaling data across teams, this episode is packed with practical insight.

🌐 More at www.dataminded.com

In this episode of The Data Playbook, we explore what it really takes to turn AI into meaningful business impact.

Host Kris Peeters talks with Joris Renkens, founder of AI product studio Guatavita, about how organizations can build AI solutions that truly work in practice.

They discuss:

• Why innovation starts with the right problem, not the latest tech
• How to validate adoption early, before writing complex models
• What makes a high-performing AI product team
• How to manage technical debt while moving fast
• Why flexible strategy matters more than fixed roadmaps

🎙 Listen & subscribe on Spotify

🌐 More at www.dataminded.com

In this episode of The Data Playbook, we explore what it really takes to build high-performance data teams.

Host Kris Peeters is joined by Rushil Daya, Senior Data Engineer at Dataminded, who shares practical lessons from years of leading successful data teams across industries.

They discuss:

• The link between data success and business value
• How mentoring beats documentation when upskilling teams
• Why testing and CI/CD matter more than flashy tools
• What makes stakeholder communication essential
• Why Agile is nothing without real feedback loops

🎙 Watch on YouTube: https://youtu.be/JEPPVakHfhA

🌐 More at www.dataminded.com

In this episode of the Data Playbook podcast, we explore what it really takes to build sustainable, data-centric organizations, moving beyond tooling and dashboards toward lasting value.

Host Kris Peeters is joined by Jonny Daenen (Knowledge Lead at Dataminded), who shares insights from years of helping organizations evolve their data strategy across sectors. Together, they discuss why data platforms, domain-owned data products, and people-first operating models are the foundations of modern data success.

🔍 Topics covered:

• Why dashboards aren’t enough for true data maturity
• From central chaos to federated teams and self-service
• The role of governance in enabling delivery
• How AI and LLMs reshape data tooling, ownership & value
• What “data-centric” really means , and why most fail to get there

🎙 Hosted by Kris Peeters

👥 With Jonny Daenen, Dataminded

🌐 Visit our website for more.

In this episode of The Data Playbook, we take a technical look at SQLMesh, a data transformation framework designed to improve the workflow and reliability of SQL-based data pipelines. Hosted by Kris Peeters, the episode features Michiel De Muynck, Senior Data Engineer at Dataminded, who provides a deep dive into SQLMesh’s internal mechanics, including its use of semantic analysis and isolated runtime environments.

Michiel outlines how SQLMesh differentiates itself from tools like dbt by incorporating a semantic parser for SQL, enabling structural validation and more precise error diagnostics during pipeline development. He also explains the implementation of virtual data environments, which allow data engineers to stage, test, and version transformations without impacting production datasets, supporting safer iteration and deployment processes.

🎧 Listen to more episodes on Spotify: Data Playbook Podcast

🌐 Visit our website for more: Website Link

In this special episode of "The Data Playbook" podcast, recorded live at the Data Mesh Live Event in Antwerp, Kris Peeters speaks with Data Mesh pioneers Jacek Majchrzak and Andrew Jones. They explore how Data Mesh addresses critical challenges in data management, including data bottlenecks, governance, and decentralization. With years of experience in the field, both Jacek and Andrew share practical lessons from their journeys and offer actionable insights into implementing Data Mesh effectively.

The conversation covers:

• Solving data bottlenecks through decentralized architectures
• Improving governance with federated models
• Aligning data strategy with business goals for impactful results
• Understanding the importance of incremental implementation
• Moving beyond "data silos" towards a more flexible, scalable approach

Jacek and Andrew provide real-world examples of how Data Mesh can transform your data infrastructure, sharing lessons on what works, what doesn’t, and how to manage a successful Data Mesh implementation. If you're looking to overcome common data management challenges like governance and scalability, this episode is packed with practical advice.

🎧 Watch the full episode on Youtube

🌐 Learn more on our website

Stay tuned for more episodes on Data Mesh and other important topics in data architecture by following "The Data Playbook" on Spotify.

Join us in this episode of The Data Playbook as we explore the sense and nonsense of data modeling with Jonas De Keuster, VP of Product at VaultSpeed. Jonas takes us through his journey in the world of data automation, discussing the role of data integration, data vaulting, and how modern data products are built using structured models. From dimensional modeling to the complexities of integrating data across multiple systems, Jonas shares practical insights into how organizations can scale their data operations.

Topics covered include:

• Data Modeling Techniques: Data Vault vs. Dimensional Modeling
• Data Automation and Integration
• Building Data Products with Scalable Models
• How to Manage Data Changes and Evolving Business Needs
• Real-world Challenges in Data Platforms

Whether you're leading a data team or just beginning your journey, this episode is a must-listen for anyone interested in the future of data architecture. Tune in for expert advice on building integrated data solutions that deliver real business value.

To learn more, visit our website, or Watch more episodes on YouTube.

What do you do when GDPR forces your cloud project to stop—and years later, you need to go back? In this episode, Niels Melotte, Data Engineer at Dataminded, unpacks the journey of a government agency that migrated from the cloud to on-prem and then back to the cloud again.

And here’s the kicker: the Big Bang migration only took 14 hours. No downtime. No data loss. No angry users.

🔍 In this episode, we discuss:

• Schrems II and why it sent European governments off the cloud

• AWS Nitro Enclaves & external key management for GDPR compliance

• Why the on-prem platform failed to meet uptime guarantees

• What “purpose-based access control” means and why it matters

• The value of standardizing with dbt and Starburst

• How data product thinking shaped the migration strategy

• Lessons learned about trust, stakeholder communication, and platform maturity

This isn’t a fluffy case study. It’s a practical guide full of engineering tradeoffs, real-world headaches, and long-term lessons. A must-listen for data leaders, engineers, architects, and anyone dealing with sensitive data and complex infrastructure decisions.

🎧 Want more episodes?Watch or Listen to all episodes of The Data Playbook on Spotify: 👉 https://open.spotify.com/show/78z3kdyBSKiURz1VnTVP9l?si=781abec722264306

Show notes, episodes & resources:👉 https://www.dataminded.com/resources/podcast

#CloudMigration #PublicSector #GDPR #DataGovernance #AWS #DataPlatform #dbt #Starburst #BigBangMigration #TheDataPlaybook #Dataminded

In this episode of The Data Playbook, we dive deep into a critical, often-overlooked question: What does it mean to build sustainable data products? And no, we’re not just talking ESG dashboards or carbon reporting.

🎙️ Host Kris Peeters is joined by Geert Verstraeten, a seasoned data scientist, founder of Python Predictions, and now a Co-Lead at The Data Forest—a consultancy that puts purpose and sustainability at the core of every data project.

Over a candid and rich conversation, Geert shares:

• How he transitioned from startup founder to corporate leader and back to startup life—with a sustainability mission.
• Why many data projects fail not because of tech, but because of missed alignment and poor adoption.
• What sustainable data products really are: tools that not only minimize environmental impact, but also stand the test of time—well-documented, actually used, and aligned with real business needs.
• Why selecting the right clients and projects is the first step toward impact, not just profit.
• How the Data Forest scores potential engagements using a unique framework: Head, Heart, and Hands.

💡 Along the way, you’ll hear thought-provoking takes on:

• The role of documentation in sustainability.
• Why good data work isn’t about building everything real-time or at scale—especially when you’re not Google.
• The paradox of GenAI and compute-heavy models in a world striving for tech responsibility.

Whether you're a data engineer, architect, scientist, or team lead, this episode challenges you to rethink what a "good" data project looks like.

👀 If you’ve ever built something technically brilliant that no one used, this episode is for you.

Hit play to hear:

• What drives Geert and his team at The Data Forest
• How to make better decisions on project scoping and client fit
• Why data professionals need to talk to stakeholders much earlier—and more often

What Not to Build with AI: Avoiding the New Technical Debt in Data-Driven Organizations

In this episode of The Data Playbook, we explore a crucial and often overlooked question in the age of generative AI: not what to build—but what not to build.

Host Kris Peeters (CEO of Dataminded) is joined by seasoned data leaders Pascal Brokmeier (Head of Engineering at Every Cure) and Tim Schröder (AI & Data Transformation Lead in Biopharma), to talk about the dark side of unlimited AI capabilities: technical debt, fragmented systems, and innovation chaos.

Topics we dive into:

• Why generative AI lowers the barrier to building—but increases long-term complexity

• The risks of treating LLMs as “magical oracles” without governance

• How RAG systems became the default architecture—and why that’s dangerous

• The zoo vs. factory dilemma: how to balance AI experimentation with structure

• Master data vs. knowledge graphs vs. embeddings – when and why each breaks down

• What Klarna got right (and wrong) by replacing SaaS tools with AI-generated internal apps

• The growing importance of AI literacy, data maps, and platform thinking

• Real-world examples of AI agents autonomously debugging systems—and when that’s terrifying

We ask tough questions like:

• Are enterprises just building themselves into a new kind of mess, faster than ever before?

• Is the AI hype driving us toward “build now, regret later”?

• Should you really let every department build their own AI stack?

Whether you're a data engineer, data architect, AI product lead, or a data strategist, this episode is a must-listen. We’re cutting through the hype to figure out where the real value is—and where the future tech debt is quietly piling up.

🧠 Key quote:"If you can't tell me why you're building it, maybe you shouldn't be building it at all."

💡 Tune in to learn how to stay smart, intentional, and strategic when it comes to building with AI.

#TheDataPlaybook #DataEngineering #AIinBusiness #TechnicalDebt #RAG #LLMs #DataStrategy #EnterpriseAI #DataGovernance #DataLeadership #KnowledgeGraphs #GenerativeAI #AIinHealthcare #AIProduct #Dataminded

What does it really take to build a modern data architecture from the ground up?

In our very first episode of The Data Playbook, host Kris Peeters, founder and CEO of Dataminded, sits down with Thorsten Foltz, seasoned data architect and engineer, to unpack what works (and what doesn’t) when designing scalable, future-proof data platforms.

With a focus on real-world tradeoffs, this episode explores:

• Cloud vs on-prem vs hybrid: how to choose the right infrastructure

• The rise of Data Mesh and when it actually makes sense

• Why fake news isn’t just a media problem—it’s a data problem inside companies too

• Vendor lock-in, cloud sovereignty, and the growing relevance of European alternatives

• The balance between open source and managed services: cost, control, and complexity

• Why team culture and communication often make or break your data strategy

• What engineers can really expect from LLMs in the data stack (spoiler: they're not replacing data modeling any time soon)

Whether you're a data engineer, architect, analyst, or tech leader, this conversation goes far beyond buzzwords. You’ll hear practical lessons, hard-earned insights, and a few uncomfortable truths about how companies actually manage data today—and how they should rethink it for tomorrow.