Chapter 034 min read
The work runs between the boxes.
Building your own AI team sounds like the safe choice. It is the slowest way to start that I know, and it puts the team in the wrong place.
Six months before anything runs
Building your own AI team sounds like the safe choice. The knowledge stays inside, and nobody else sees the data. It is also the safest slowest way to start that I know.
Month one is a job ad. Month three, your favourite candidate takes another offer. Month five, the team has started and is waiting for access to the systems while the security review runs. All told, building in‑house costs easily over €100,000, and it takes about six months before a first system runs.
An ambassador without an assignment
Many organisations start smaller than a team. They appoint an AI ambassador. One executive told us about theirs with a shrug: the ambassador, he said, does not do much. Then he gave the reason, and I have repeated it many times since. An ambassador without an assignment you can measure is pointless.
The other version is the lone enthusiast. One person becomes very good with AI, builds remarkable things, and ends up isolated from the rest of the organisation, with tools and habits nobody else shares. A founder who sells AI tools told us he had lost customers exactly that way.
A team in a box improves the box
Put the team under IT and it fixes your problems IT’s problems. Put it under sales and it fixes sales. That is no criticism of the team. It is what a place in the organisation does to anyone.
But the hours are rarely lost inside a department. They are lost between departments: where sales hands over to planning, and planning to finance. As one executive put it to us: IT owns the tools, finance owns finance, sales owns sales, and nobody owns the process that runs through all three.
There is a second reason. People who join an organisation learn how it works by taking on its habits. That is the opposite of asking why the work runs this way at all. An outsider asks that question in the first week, because nobody has told them the answer yet.
Know the work, then hire
I would turn the order round. First find out how the organisation operates, and which processes cost the most. Rebuild one, and see what it is worth. Then you know which processes deserve a team, and what those people should be able to do.
If an organisation hires its own team after working with us, good. Now it knows what to hire for.
When building it in‑house is the better choice.
- You already have an AI team with time to spare.
- Your processes are your product, and the know-how must sit with your own people.
- You plan years ahead and can wait for the first result.
Then build it in‑house. We’ll say so in the first call.
Questions
What teams ask when they compare.
Can we do both?
Yes, and many organisations should. Start with an audit, and let your own team take over the workflows we rebuild. Everything runs in Grasp Work with an owner on your side, so nothing depends on us.
Does our IT team stay in control?
Yes. Grasp Work connects to your tools with the permissions your IT team approves, each organisation gets its own deployment, and you can bring your own model keys.
What happens to the workflows if we stop working with you?
They keep running in Grasp Work, with their owners on your side and their measurements intact. You decide whether we rebuild more.