Which AI Model Powers This, and What If It Changes?

DD Pim van den Boomen • 18 July 2026 •5 min

Pillar 02 // The catch

model.md

$ which model --and-what-if

Which AI model powers this, and what if it changes?

Claude, from Anthropic. And yes, that can change: this field moves faster than any contract. If something better comes along, we switch, and that's work we do, not you. What you need to know is which part of your system moves along and which part gets rebuilt. That's the honest version of "no vendor lock-in".

DD DataDrift Digital • July 18, 2026 • 6 min

This question almost always comes from someone who's already been stuck with a tool. It's not a technical question but a risk question: if I buy this, what exactly am I locked into.

The answer below is more precise than "you're not locked into anything", because that promise can't be kept. There is a boundary, and it runs in a place you can know in advance.

01 / the answerWhich AI model do you use?

Claude, from Anthropic. Within that family we pick a different model per task: heavier models where real reasoning is needed, lighter and faster ones for recognizing and sorting. That choice sits under the hood and changes when something better comes out.

That's the factual state. The more interesting half of the question is what happens when that changes, and that comes further down.

02 / the choiceWhy Claude and not something else?

Three reasons, in the order in which they weigh for us.

Your data doesn't go into the model. For business use through the API, the standard is that what goes in isn't used to train the model. For a system that works on your customer data, that's not a detail but a condition.
It reasons best over multi-step work. That's our own finding from our own build practice, not a leaderboard. On a task like "read this conversation, pull out the agreements, put them in the system, and flag the exception", the difference between models is large.
The vendor is accountable on safety. Anthropic publishes where the model's limits are and what it shouldn't do. We'd rather build on a foundation whose builder writes down its own flaws.

What's not said here: that Claude is the best model in the world. We don't know that and it changes by the month. It's the model that currently fits best with what we build.

03 / the switchWhat if there's a better model tomorrow?

Then we switch. That's not an exception but part of the maintenance you pay for monthly: model changes and connection adjustments are included. You get no invoice for it and you don't have to do anything.

Honestly, though: switching is work, not a toggle. Every model responds differently to the same instruction. An instruction that does exactly the right thing with one model becomes too formal, too elaborate, or slightly too creative with another. That means retesting and adjusting, per role, on your material.

The model is interchangeable. Tuning it to your work is not.

In practice we therefore don't switch with every new model, but when it's measurably better for a task you actually use. Curiosity is not a reason to open up a working system.

04 / the boundarySo am I locked into one vendor?

Partly, and this is where the boundary runs. Your system consists of three layers that don't move house equally easily.

MOVES WITH YOUThe knowledge layer. Your way of working, your offer, your examples, your phrasing: that's text and structure. Fully model-independent and yours. If you switched to another model or another provider tomorrow, this would simply come along.

PARTIALLYThe connections. The links to your email, your calendar and your admin systems hang off those systems, not off the model. They keep working, but the pieces that process the model's output need adjustment.

REBUILTThe tuning. The instructions behind each role are tuned to one model. That's the part that gets recalibrated on a switch, and it's exactly where our hours go.

Anyone claiming zero dependency is selling you the first layer and staying quiet about the third. The layer that matters in an actual break is the first, and that one is yours and portable. That's what we can promise.

05 / the alternativeWhy not a custom or open model?

It's possible, and for some businesses it's the right choice. Putting a publicly available model on your own infrastructure gives you the most control. It also costs the most: hardware or server costs, someone to maintain it, and less reasoning power per euro than what you get from a major provider.

For a company with a handful of people, that rarely pays off. If you ask about it specifically because there's a reason your situation is different, that's a conversation we take seriously and don't wave away.

06 / usableWhat should you ask any provider about this?

Which model, and who decides when it changes: you or me?
Is my input used to train the model? Ask for the answer in writing.
What can I export if I want to leave, and in what format? No format means no answer.
What does a model switch cost me? If the answer is "nothing", ask who does the work.

Frequently asked questions

Which AI model do you use?+

Claude, from Anthropic. Within that family we pick per task: heavier models where reasoning is needed, lighter ones for sorting and recognizing. That choice changes when something measurably better comes out for a task you actually use.

What happens if a better AI model comes out?+

Then we switch, and that's included in the monthly maintenance. You don't get a separate invoice for it. Switching is work, though: every model responds differently to the same instruction, so the tuning per role gets recalibrated and tested on your material.

Is my company data used to train the AI model?+

No. For business use through the API, the standard is that input isn't used to train models. That was a condition for us in choosing, not an afterthought. Check this with every provider and ask for the answer in writing.

Am I locked into one AI vendor if I buy this?+

Not for the most important part. Your knowledge layer, your documented way of working and phrasing, is text and moves with you everywhere. The connections need adjustment, and the instructions behind the roles are tuned to one model and get recalibrated on a switch. Zero dependency doesn't exist; what matters is which layer is yours.

Can you also work with an open source model or ChatGPT?+

Technically, yes. An open model on your own infrastructure gives the most control and costs the most in hardware, maintenance, and reasoning power per euro. For a company with a handful of people that rarely pays off, but if there's a concrete reason your situation is different, we'll have that conversation.

30 minutes, free, no sales pitch

Want to know where the dependency would sit for you?

Schedule a call

We look at where your time leaks away and tell you honestly whether we can do something about it. Often the answer is: you can do this yourself. Then we say so, and you've spent thirty minutes on a clear answer.

→ Schedule a call (30 min)

This text was produced with AI assistance and reviewed and approved by a human before publication.