Fixed price
€7,500
- Four sessions of about two hours
- Two weeks, run over video
- A written AI integration plan
- Not useful after session one? You pay nothing.

At division5 we went from people running every process to more than 85% of them automated, and improving a process went from months to weeks. It did not start with AI. It started with listing the work, putting the data in one place and building one system, and only then adding agents and workflows.
The free guide sets out that order. This sprint is for the companies that want to go through it with someone who has already done it, for their own business.
It is a new offer, so there are no client stories here yet. The proof is division5 itself, and the promise below: if the first session is not useful, we stop and you pay nothing.
How the two weeks run
Four sessions, with real work in between
Four sessions of about two hours, twice a week. The gaps are where your team checks the real processes and the real data, rather than describing them from memory.
01 · Session one
List the processes and where the hours go
Every recurring process across sales, marketing, operations, HR and finance, who does it today and what it costs in hours. We pick the few where AI pays back first. For a service business this session also covers what you actually sell and which parts repeat, so the AI goes onto work that repeats rather than work reinvented for every client.
Before the next one: Each owner lists the data their processes depend on, and where it lives today.
02 · Session two
Map the data
Where each kind of record really lives, and where there are two answers. This decides what has to be fixed before any AI goes in, and whether the system your team works in can carry it.
Before the next one: List the decisions in those processes that follow a fixed rule every time, with no exceptions.
03 · Session three
Design the agents
For each chosen process, which steps an agent reads, drafts and proposes, and who approves its work. The rule is the one we run on ourselves: AI proposes, a person decides.
Before the next one: Name the person who will approve each agent, and check they have the time.
04 · Session four
Decide the order
What becomes an automated workflow, what stays with an agent and a person, and the ninety days that get the first process running. With owners, and small enough to actually happen.
What you leave with
A plan you can build, in order
A map of your processes, with the ones where AI pays back first marked
A data map: where each record lives, and what to fix before anything else
The first agents: what each one proposes, and who approves it
A ninety day plan with owners and an order, including what it takes to keep it running
€7,500
Fixed, for all four sessions and the written plan afterwards. Run over video, so your timezone matters more than your address. If you decide after the first session that this is not going to be useful, we stop and you pay nothing. The plan is yours to build with whoever you choose, including divisionAI, which builds these systems.
Who this is not for
Four cases where I would tell you not to book
- Anyone who wants a tool recommendation. I will not tell you which AI product to buy
- Companies that want a demo for the board rather than a change in how the work gets done
- Teams whose leadership will not clear four two-hour blocks
- Anyone who wants AI making decisions with no person checking them
Run a service business? The sprint starts with productizing what you sell, because AI on top of work that is reinvented for every client only makes the mess run faster. The free Productize and Scale playbook is the do-it-yourself version of that part.
Book
Tell me about your business
What you sell, roughly how many people you are, and where the hours go. A paragraph is enough. I will reply with whether I think a sprint is the right thing, and say so plainly when it is not.