Expertise rarely sits in the tool – it sits in the understanding of data
If you’ve spent a few years working with data and moved between different environments, you’ll know that most things are built on the same underlying principles. The syntax may differ. Interfaces may look different. But the logic remains the same. Switching ETL tools or platforms is rarely what determines whether you can deliver value.
What truly matters is whether you understand your data:
- How do data flows connect?
- What defines a business key?
- How should historical data be managed?
- How should models be designed?
- How does data quality impact the entire chain?
That’s where the real expertise lies. I often compare it to changing cars – different brands place controls in different locations, some even have the gear lever on the steering column, and if you hire a car in the UK, everything feels reversed. Yet most drivers manage just fine. You’re likely just as good (or bad) a driver in a Skoda as you are in a Mercedes.
AI and AI-assisted development do not change this
There’s a lot of talk right now about AI, AI-assisted development, and the idea that coding is becoming “free” – that technical craftsmanship will no longer be something you can charge for. I believe that’s a dangerous oversimplification.
AI can absolutely generate code. It can create pipelines, transformations and data models in seconds. But if the person using these tools doesn’t understand the data, the outcome can be unpredictable.
AI doesn’t know which business key is actually correct.
It doesn’t know how historical data should be handled.
It doesn’t know what regulatory requirements apply.
It doesn’t know how different data sources should be harmonised.
It doesn’t know how the architecture needs to be designed to last over time.
All of this still requires experience and understanding. And frankly, coding has never been the hardest part of our job. The real challenge has always been understanding the business, the data, and how that data needs to be treated in each specific context.
We use AI-assisted development – grounded in data understanding
Of course, we also work with AI-assisted development and closely follow its evolution. We test different tools, discuss what works well and what doesn’t, and continuously explore how to apply them effectively in our clients’ environments.
But AI is a tool – not a replacement for expertise.
The easier it becomes to generate code, the more important it is that someone actually knows what should be built, why it should be built, and how the data needs to be structured for the solution to work over time.
