Claude Partner Network · AI Strategy & Roadmap
An enterprise AI strategy is four decisions with named owners, not a forty-page deck.
Two weeks, fixed price, and the deliverable is a decision record your own people can edit after we leave.
Pexon runs enterprise AI strategy as a two-week engagement ending in four decisions, each with a named owner: which use cases, in which order, under whose permissions, against which budget. The deliverable is a scored portfolio and a costed roadmap rather than a slide deck. Fixed price, €4,900, net of VAT.
Where you are
Five rungs, and you buy the first one
- Where most companies areEnthusiasm with no ownerSeveral teams are using Claude on their own files, nobody has written down who approved it, and the first invoice arrived as one line.
- Step oneA portfolio with the rejections in itCandidate use cases scored against the same criteria, and the disqualified ones kept on the page with the reason. A list of only the survivors cannot be audited.
- Step twoThree decisions with three namesIdentity, tenancy and budget each resolved to a group, an account and a cost centre, each agreed by the person who owns it.
- Step threeA sequence with dependencies markedThe order the use cases get built in, and which ones cannot start until a connector or a permission model exists.
- Step fourA cost model against your own volumePublished token prices applied to your expected traffic, with the caching and batching assumptions stated so the number can be challenged.
The portfolio
Scoring a use case, including the ones that fail
What puts a use case on the roadmap
- A named owner. One person whose job gets easier or harder depending on whether it works. Not a department.
- A baseline we can measure. How long the task takes today, how often it is done, and how often it is done wrong. Measured in your first two weeks, not assumed.
- Reachable data. The system holding the answer has an interface, or building one is itself on the roadmap with a date.
- A tolerable failure. What happens when the model is confidently wrong, and whether a human sees it before the customer does.
What takes one off
- The permission model is the use case. If answering the question correctly requires an entitlement decision nobody has made, the entitlement work is the project and the assistant is the easy part.
- The process is broken underneath. Automating a process whose rules nobody can state makes it fail faster and less visibly. That one goes to process redesign first.
- The volume does not justify it. A task done four times a month by one person is a bad first use case however well it demos.
- Nobody will own the output. If no team will take responsibility for what the system produces, it will be switched off within two quarters whatever it cost.
The two weeks
What happens on which day
Days 1 to 3
Interviews and the current bill
Three to six interviews, plus whatever the model already costs you. The existing spend is usually the most informative document in the company and almost never the one offered first.
Days 4 to 6
Candidate use cases, scored
Every candidate against the same criteria, including the ones that fail. The rejection list is a deliverable, not a by-product.
Days 7 to 8
The three decisions
One session with the identity owner, the cloud owner and the budget holder in the same room. This is the part that cannot be done asynchronously, and the part that most often has to be scheduled twice.
Days 9 to 10
The cost model
Published token prices applied to your expected volume, with the caching and batching assumptions written down so your own finance team can argue with them.
Week 3
Handover and challenge
We present it, your people try to break it, and the roadmap changes in the room. A roadmap that survives its own handover unchanged was not read.
What you receive
Five artefacts for €4,900, net of VAT
- The scored portfolio. Every candidate use case, its score against the published criteria, and for each rejection the reason it was dropped.
- The decision record. Identity, tenancy and budget resolved to a named IdP group, a named cloud account and a named cost centre, with the person who agreed each one.
- The sequenced roadmap. Build order with dependencies marked, so it is visible which use cases are blocked on a connector or a permission model rather than on capacity.
- The cost model. A spreadsheet, not a picture of one: published token prices against your expected volume, with every assumption in a cell you can change.
- The things we told you not to do. Written down with the reasoning, because the next person to propose one of them will arrive in six months and will not have been in the room.
When we are the wrong choice
Three cases, and the first is the common one. If you already know your first use case and the person who owns it, this engagement confirms an afternoon's thinking over two weeks and you should buy the build instead. If what you need is a board paper rather than an engineering decision record, a strategy house will write a better one than we will, and that is a real difference in product rather than modesty. And if the honest blocker is that nobody senior has decided AI is worth doing, no roadmap fixes that; the engagement will produce a good document that sits unread, and we have watched it happen.
Pexon Group is a Certified partner in Anthropic's Claude Partner Network (Services Track), granted on 1 October 2026. This site is published by PEXON ROMANIA S.R.L., a separate legal entity in the same group, and the party you contract with here.
Where this goes next
- The Claude Partner Network catalogue — All sixteen Services Track categories and which page each one lands on.
- AI proof of concept — If the roadmap says build, this is the decision the POC brief has to make first.
- EU AI Act readiness — The readiness and governance assessment, in the form it takes for a European deployment.
- What Claude costs — The four things that drive the bill, which is what the cost model in week two is built on.
What a sponsor asks before approving this
How is this different from a consultancy's AI strategy deck?
The test we hold ourselves to is whether the document names a person next to every decision. A deck that says identity, tenancy and budget must be resolved is describing the problem. A roadmap that says which IAM group, which cloud account and which cost centre, with the three names of the people who agreed, is a decision record. The second one is two weeks of work; the first one is a week of writing.
What do we actually get at the end of the two weeks?
A scored use-case portfolio with the disqualified ones listed and the reason each was dropped, a sequenced roadmap with the dependencies marked, the permission and tenancy decision written down, and a cost model built against published token prices and your own expected volume. Everything is in a format your own people can edit after we leave.
Can you do this if we have not chosen a model or a platform yet?
Yes, and that is the better time. The platform decision changes the cost model and the residency answer, so making it inside the engagement rather than before it means the roadmap is built on the platform you chose for a reason. If you have already standardised on Bedrock or on Azure, that shortens the work rather than changing it.
Who from our side needs to be in the room?
Three people, and the engagement does not work without them: someone who can commit on identity, someone who can commit on the cloud tenant, and the budget holder. Everyone else is optional. The reason pilots stall is that those three have never been in the same meeting, so getting them there is part of what is being bought.
What does it cost, and what is not included?
€4,900 fixed for the two weeks, net of VAT. Not included: any build. Model consumption is billed to you directly by Anthropic, AWS or Microsoft at their published rates — we do not resell tokens. If the roadmap says the first use case is worth building, that is the pilot engagement from €15,000 and it is a separate decision you make afterwards.
When is a strategy engagement the wrong purchase?
When you already know the first use case and who owns it. Then this is two weeks of confirming what you could have confirmed in an afternoon, and the money is better spent on the build. Say so on the call and we will tell you the same thing — a roadmap for a company with one obvious use case is an expensive way to write it down.
Next step
Bring the three people who can decide
Thirty minutes with the architect who would run the engagement. If you already know your first use case and who owns it, we will tell you to skip this and spend the €4,900 on the build instead.