For years, the difference between low-code and full code was clear. Low-code was fast and affordable, but you hit its limits. Full code could do almost anything, but you paid for that with long projects and hefty budgets.
AI has turned that upside down. Low-code now lets you do far more than before. And full code has become at least as fast.
So what is left of low-code? Elianne Burgers (Team Lead Low-code and AI Solutions) had her answer ready:
"I think the term low-code will disappear. But a platform where we can offer security and stability, that will stay. It no longer matters how you built it."
In this article:
- Clients no longer come with an idea, they come with an app
- The platform handles what you would otherwise have to figure out yourself
- AI finally makes sure agreements are followed
- Without oversight, AI builds a table that already exists
Clients no longer come with an idea, they come with an app
Clients used to come to Elianne with an idea. Now they come with something they vibecoded themselves.
"Clients now come to us with what they built themselves. With the question: we built this, now what?"
That gets her excited. Anyone who builds something themselves has already translated their business process into software. Including all the jargon and company-specific knowledge that comes with it, which is often underestimated. The conversation starts at a different level. You could hardly ask for a better briefing.
But an app that works for one person is not yet running across the organisation. Elianne draws the line at the moment you start sharing it. Building something for yourself is great. Once you share it with your team or the whole organisation, she says: "Let's sit down together first." Then it's about testing, security, compliance and a mature path from development to production. Topics the builder has rarely thought about.
The platform handles what you would otherwise have to figure out yourself
If AI does the building, a platform has to offer something other than speed. Richard Wierenga (Tech Lead Low-code) keeps it practical.
"We use AI to create the solution, but the platform gives us a place for it to land. We don't have to think about where it will be hosted or where our database will live."
No spinning up a web server. No configuring a firewall. Roles and permissions an organisation has already worked out simply keep applying. And company data doesn't quietly end up on social media. Put your customer data and an external channel in different groups, and the platform blocks any app that connects them. That is not an agreement on paper.
It isn't free. You pay for licences, and for a single solution that feels like a sizeable investment. Elianne then looks at the bigger picture: with the same licence you will soon cover several business processes. Richard points to the other side. If you build it yourself in full code, you need people to handle the management.
AI finally makes sure agreements are followed
Every organisation has rules for how you build. Who checks whether they are followed?
Elianne: "We used to have naming conventions. You wrote them down in a manual somewhere. Whether they were followed, we often didn't check. And when we did check, we were often disappointed."
Familiar to anyone who has ever written a policy document.
Now you set up those rules with the help of AI, and AI also helps you follow them. The team captures everything in a second brain: platform knowledge, governance, naming rules and which data someone is allowed to use. Add to that transcripts of client conversations and research into the client's industry. All in one place, structured in one way. Not in someone's inbox or on a loose scrap of paper.
Richard talks about building skills. Knowledge in his head for analysing an application or reviewing an architecture choice, he captures in a skill. That way the whole team can use it. It also helps with existing apps that have no documentation. A skill documents the application and looks at what can be improved.
The effect is concrete. When someone takes over a project from a colleague, they get up to speed much faster. Everyone works with the same set of rules.
Without oversight, AI builds a table that already exists
You might think those skills can finish the job. Not quite.
Richard gave an example. Power Platform has standard tables, such as accounts. If the client calls that table "customers", AI sometimes creates a new table. Even though it was already there.
Elianne sees the same on the functional side. AI sometimes comes up with ideas where you think: that might have been true five years ago. Experience and keeping up with what changes are still needed.
Documentation shows clearly how that division works. Nobody likes documenting, and automating it saves a lot of time. The quality is often better too. But then you're at 80%. A human dots the i's.
Conclusion
Will low-code still exist in three years? Maybe not as a term. The promise behind it will: letting people in your organisation build software themselves. How you build something matters less and less. Where it lands matters all the more.
Want to read more? In the white paper Low-code and AI (in Dutch), Elianne and Richard go deeper into governance, agents and what changes for Power Platform users.
The full episode is now available to watch and listen to (in Dutch)!
FAQ
Frequently asked questions
We'd be happy to answer those right away. If your question isn't listed, please feel free to contact us.
