The short answer
Hiring an AI developer starts with what the work is. If you want to build AI into your product or processes, such as an agent, a chatbot on your own documents, or an integration with a language model, you need an AI engineer: a developer who builds software around language models. If you mostly want your software finished faster, you need a senior developer who uses AI while writing it. Training your own model is the third, rare case.
Then you check one thing: has this person put AI into production, or only built a demo? Anyone can build a demo in an afternoon. The work is in what comes after.
Which AI developer do you need?
“AI developer” covers three very different roles. Hire the wrong one and you pay for expertise the project never uses.
| AI engineer | Developer who uses AI | ML specialist | |
|---|---|---|---|
| Builds | Software around existing language models: agents, chat on your own data, integrations | Regular software, faster, with AI helping to write it | Custom models, trained on your own data |
| Choose when | AI becomes part of your product or process | You want an app, platform, or integration built | No existing model solves your problem |
| Must know | Language models, tools, evaluation, security, and cost | Architecture, review, and testing | Statistics, datasets, and training |
| How often needed | More and more | On almost every software project | Rarely, with large amounts of your own data |
Most companies looking for an “AI developer” today need the first one. Training your own model is rarely the cheapest route: the models from Anthropic, OpenAI, and Google already handle most tasks when you use them well.
Demo or production: the difference starts after week one
A chatbot that answers questions about your own documents takes a day to build. That’s exactly why a demo says so little. The work is in what the demo skips:
Measuring whether it's right
A language model always gives an answer, even when it's wrong. A good AI engineer builds a set of test questions with known good answers and measures whether each change makes things better or worse.
Limits on what it may do
An agent that can send email or write to your systems needs limits. What can it do on its own, where does a human sign off, and what happens when someone tries to trick it?
What it costs per month
Every answer costs money per word processed. What costs cents in a demo can become a real bill at a thousand users a day. An experienced engineer works that out up front.
What happens when it fails
The model is slow, the provider is down, or the answer makes no sense. In production you have to see that and handle it before a customer notices.
A demo shows it can work. Production shows it works well every day, even when nobody is watching.
Five questions for your first conversation
You don’t need AI expertise to assess an AI developer. Ask these questions and listen for concrete answers:
Which of your AI systems is running with real users right now?
Ask about something that's live, not a prototype. How many users, for how long, and what went wrong?
How do you know the model answers well?
A good answer mentions a fixed set of test questions and a way to measure. "I try it out" is the answer of a demo builder.
What will this cost per month once it's live?
Anyone who has worked in production has seen a bill and can estimate one from the number of users.
What should the AI not be allowed to do?
Someone who gives examples right away has thought about security and about what a model does with your data.
When would you not use AI?
A good AI engineer also tells you when a plain line of code is better, cheaper, and more reliable.
Freelancer, agency, or in-house?
An in-house AI engineer is scarce and expensive, and rarely needed for a single project. A freelancer is available fast, but you’re on your own if they leave or get stuck. An agency or collective gives you several senior people who review each other’s work, and someone to finish the project when things get hard.
Which fits depends on how long the work lasts. For a well-defined AI project, a fixed price usually works best. For a product that keeps growing, a fixed number of hours per month. How hiring works with us is on the software development support page.
What does an AI developer cost?
In the Netherlands, a senior AI engineer charges about the same as other senior developers, roughly 70 to 125 euros per hour excluding VAT. With us, AI work costs the same as other development work: 100 to 150 euros per hour.
On top of that, count the cost of the model itself. You pay the provider per use, separately from the developer’s hours.
Conclusion: hire the role, check the work after the demo
First decide whether you want to build AI into your product or mostly want software faster. Then find someone who can show what happened after their demo went live. You won’t see that difference on a CV, but you will in the answers to five questions.
How do I hire an AI developer?
What's the difference between an AI engineer and an ML engineer?
How much does an AI developer cost per hour?
Do I need an AI developer or a regular developer?
How do you recognize a good AI developer?
AI in production, not just in a demo?
We’re senior developers with 15+ years of experience who build AI agents and integrations that run every day, with tests, limits, and a predictable bill.
