Managing your AI tooling
Your people are building things with AI themselves. We make sure it can go live safely — even when we did not build it.

What we take over, and what you keep
Five parts. Some run continuously, others come round at set moments — and for all five it holds that we need not have built it.
How often
What
What that means
On takeover
01
Taking over what already runs
Even if we did not build it. Which sites, integrations and agents are running, whose account the hosting sits on, which domains expire and who still has access. Rarely a pretty list, and that is exactly why it starts there.
Before go-live
02
Making it safe before it goes live
Access, keys, and which sources a model may see and where that data goes. With something put together quickly this is almost always why it is not live yet — and usually it is sorted within a week.
On paper
03
A team behind you, with response times on paper
What counts as an outage, how fast we respond and who picks up. On paper, so you can show it to your own client. It also says what is not included.
Continuous
04
Updates, backups and a rehearsed restore
Updates on a fixed cycle, regardless of whether there happens to be time. Backups kept apart from the server they came from. And we restore one onto a test environment once: a backup you have never tested is an assumption.
Always
05
You keep building
No lock goes on the door. Your people keep access and keep building; we look over what goes live. Want to manage it yourself later, then we hand it over with the documentation.

What we work with
- ShopifyPartner
- StoryblokPartner
- VercelPartner
- GooglePartner
- MetaBusiness Partner
- SanityPlatform
- ClaudeTooling
- ChatGPTTooling
- GitHubTooling
- NotionTooling
What we have built
Some we built ourselves. Others we took over.
01 / 03
Common questions
Not answered here? Ask the AI in the top right, or call us.
No, but we look first. We take over management of sites we did not build, on the condition that we may look inside before we promise anything. Sometimes the technology turns out to be so old that managing it costs more than replacing it. Then replacing it is our advice, even when that means you do not become a client yet.
That depends on what you agreed. Which is why response times are on paper instead of in a promise. For a store in peak season you agree something different than for a company site, and you pay something different for it. What does not change: you get someone who knows the system, rather than a first line that raises a ticket.
Yes, and that is the starting point. We do not put a fence around what runs: your team keeps access and keeps building. What we add is the agreement about what happens before something goes out — which data may leave, who switches it on, and who gets called when it falls over. That is less strict than it sounds, and it is usually the only thing missing.
It depends on what is running and how fast you want an answer. A company site on a monthly cycle sits low; a platform with integrations carrying your revenue is a different conversation. We charge per month, because per incident we would earn from your outage, and that is a strange incentive.
Often yes, and the first question is not a technical one: which sources does it read, where does that data go and who may change it. With something put together quickly it is usually not written down anywhere, and then that is where we start. Only then do we look at the hosting.
Is it live yet?
The biggest risk is not what you build, but what you therefore never switch on.
Tell us what your team built and why it has not gone out yet. In half an hour you hear what it takes to switch it on, and roughly what that costs. Even if we do not end up managing it.





