At Innovation Day: AI Made Real 2026, Akshay Jadhav and Jorrit van de Geer revealed how Albert Heijn has been advancing GenAI inside one of Europe’s most complex retail environments. Their talk presented a clear story of early experimentation, steady platform maturation and fast‑growing organizational adoption.
It also highlighted a new engineering practice that brings human expertise and AI agents together to speed up software creation across the company. The session offered a grounded perspective on how a large retail organization turns AI from experimentation into everyday capability.
Building a secure, scalable foundation for GenAI
Akshay opened by explaining how Albert Heijn first started to experiment with GenAI in 2023, when internal hackathons gave teams the chance to spot new opportunities. This led to the creation of AI Labs, which eventually became the GenAI Platform, created in collaboration with EPAM. The GenAI Platform is a secure hub where teams can tap into powerful AI models through easy-to-use APIs.
The team introduced cost controls, load management, model governance and clear guidelines so that the rest of the organization could experiment with confidence. These measures created an environment where developers and non‑technical teams could innovate without compromising security or compliance.
“When people understand what AI can do for their work, adoption grows naturally.
Akshay Jadhav, Head of Technology at Albert Heijn Stores Operations
As adoption grew, the platform team added accelerators, templates and hands‑on consulting. With these new tools in place, teams could quickly turn their ideas into real prototypes, all while sticking to best practices. Now, Albert Heijn has dozens of AI use cases up and running, and the platform is powering billions of tokens every month.
Supporting customers and empowering store teams
The platform quickly enabled high‑impact use cases. One well‑known example is Steijn, the personal dinner coach in the AH app. It chats with you to suggest recipes, tailor meal ideas to your needs, and even help you use what’s already in your fridge.
Another key initiative is the employee assistant, which empowers 100,000+ employees with instant answers on product, price and stock questions. This assistant makes store tasks a breeze and matches what younger team members expect from their tech.
The platform also directly supports developers. Langflow became an internal environment where technical and semi‑technical teams can design AI agents in a controlled setting. This helped spread AI literacy across the organization and encouraged more experimentation.
Addressing the challenge of modernizing retail systems
Jorrit then described a problem that many large retailers will know all too well. Some store applications are old, lack documentation or were created in tools that are difficult to extend. Rewriting these systems takes time and requires a deep understanding of how current versions behave. One example is Schoon, an application that handles hygiene and quality assurance management in stores that needs a modern redesign.
At the same time, AI models for coding and analysis have improved quickly. Tools like GitHub Copilot and specialized coding agents can already generate, interpret or refine code with high accuracy. This inspired the Albert Heijn team to explore how AI could support every stage of the software development lifecycle.
Structured AI‑assisted engineering
This approach uses a loop between engineers and AI agents. The engineer creates a focused prompt with clear intent, context and constraints. The agent generates code or architectural components. The engineer reviews the output, adjusts it where necessary, and updates the agent’s instructions. Over time, the agent becomes better aligned with the team’s standards. The work becomes faster, and the engineer remains in control of quality and design choices.
“Clear instructions create better AI output. Precision is what turns experimentation into reliable software.”
Jorrit van de Geer, Transformation manager at Albert Heijn
Jorrit compared this to guiding a construction contractor. If instructions are vague, the result may look acceptable but will not match the specifications. If instructions are precise and delivered in the correct sequence, the contractor produces something reliable and future‑proof. Structured AI‑assisted engineering follows that same principle. The team scopes work into smaller parts, sets clear requirements and builds each component step by step.
The results have been measurable. The team now develops twice as fast as before and continues to gain speed as engineers refine their agents.
What other organizations can learn from this approach
Akshay and Jorrit closed with several observations that can guide any organization working toward responsible and effective AI adoption.
- Create a strong starting point. Security, compliance and access governance ensure that experimentation can happen in a safe and responsible manner.
- Support the right mindset. Leaders should encourage teams to view AI as a tool that expands what they can achieve.
- Build a team that mixes senior expertise with curiosity. Experienced engineers are essential because they can assess AI‑generated code and guide agents toward better output. As Jorrit said: “Curiosity and senior expertise create the strongest partnership with AI.”
- Encourage experimentation across the organization. Workshops, tech talks and internal exploration sessions help people gain confidence and uncover new possibilities.
- Bring in partners who offer practical experience. Working with teams that have solved similar problems accelerates progress and helps avoid early pitfalls.
A new pace for enterprise engineering
The work that Akshay and Jorrit shared shows how GenAI becomes part of everyday engineering practice. Their approach supports faster delivery, clearer architecture and better use of human expertise. As more teams adopt structured AI‑assisted engineering, Albert Heijn gains a stronger rhythm where exploration and disciplined execution move together. The progress is already clear from the dry-run session, showing that this model can keep shaping how digital products evolve at Albert Heijn.
Curious how this applies to your organization?
At Empathy Lab, we work with teams to explore what orchestration, AI readiness, and experience design mean in practice, across technology, language, and organizational structures.
If these questions are part of your current challenges, we would love to continue the conversation.
Contributor in this article
