Home · Services · AI strategy & roadmap
Service · AI strategy & roadmapThis one isn't software. It's the person, in the room.
Four of our six services end with software running on your systems. Two don't — this one and AI enablement. Here you are working directly with the people who would build it — mapping where AI actually fits your business, with the same hands that would do the building if you decide anything should be.
In the roomInterviews and working sessions — nothing is installed, and nothing of ours is running when we finish
Thirty minutes, no obligation, and nothing is pitched on it. The call is with the people who would do the mapping — and if the honest answer is that you don't need a roadmap yet, that is what you hear on it.
Interviews, working sessions, and a plan that survives argument.
There is no product here. The work is listening to the people who do the job, putting every idea on the same map, and writing down what to do first — with the reasoning attached, so your leadership can argue with it instead of nodding at it.
You leave with a decision you can defend.
Four things this engagement is scoped to produce, whatever your business turns out to need — and then the paperwork that outlives it.
A clear-eyed view of where AI can carry real weight in your business — and where it would only waste effort. The second list is the one that stops effort going where it would not come back.
A staged plan: the win you can see quickly, then the bigger move behind it. Each step sized by what difference it would make against what it would take.
Who decides, which data is ready, what needs protecting — written the way your team speaks, so the rules survive contact with a busy week.
Priorities, trade-offs and reasoning in business terms — so leadership can challenge the plan, and then own it.
On paper, and yours to hand to anyone
In the document
- An opportunity mapevery AI use case you brought and every one we found, scored for impact and for feasibility against the systems you actually run
- A phased roadmapwhat to do first, next and later — with owners and milestones, not just ambitions
- A solution-development roadmaphow the solutions you choose get designed, built and rolled out, phase by phase, AI or not
- A data-readiness checkwhat your systems can support today, and what has to be fixed before anything else can be attempted
- Tool and model recommendationswith the reasons attached — never brand loyalty, and never a reseller's margin
- A skip listthe ideas that are not right yet, and a written reason for every "not now"
- A plain-language summaryshort enough to send to your whole team, and written so it is read rather than filed
How the work is done
- Built from interviews with the people who do the worknot just the leadership view of it
- Working sessions, not slide dumpsthe decisions get made in the room, while the people who have to live with them are still in it
- Every recommendation explained until it is understoodagreement that came from following the argument, not from trusting the consultant
Every recommendation is a decision you make, not one we make for you.
A roadmap that hands you conclusions is a roadmap you cannot argue with. This one hands you the reasoning and the consequences of being wrong, and leaves the call where it belongs.
- IFa recommendation would commit your business, change someone's job, or move data out of your systems
- ORthe evidence behind it is a claim rather than something we watched or measured
- ORthe people who would live with it have not been asked
- THENit goes on the map as your decision, with the reasoning and the consequences of being wrong written beside it
Nothing on the roadmap becomes a commitment until you make it one. Stop after the map and the map still stands on its own — it is written to be useful to whoever does the work next, including someone who isn't us.
Five phases — and on this one, the build is a plan.
The order never changes. What changes here is phase three: it produces a document rather than a running system. The doors are in the same places — you can stop after any phase and keep what has been made by then.
Listen
A 30-minute call about your goals and your pressures, then short interviews across your team — the people doing the job, not only the people describing it.
Map
A working session where every idea lands on the same map: what difference it would make against whether your systems could carry it. Scored in front of you, not delivered later.
Build
The roadmap gets drafted — the order, the owners, the milestones, the skip list — with the data-readiness check sitting underneath all of it.
Prove
We pressure-test it with you until it survives honest questions. A plan nobody has argued with has not been tested, it has only been presented.
Hand over
The decision session. You leave knowing what to do first, what to skip and why — in writing, in your words, and yours to take anywhere.
No lock-in by design — and no obligation to build what the map recommends, with us or with anyone else.
Where a roadmap is the wrong move.
The fastest way to waste a strategy engagement is to start one when the answer is already obvious. Both lists below are policy rather than a pitch — they are printed so a reader can talk themselves out of this engagement before committing to anything.
What a roadmap engagement is not
- Not software. Nothing is installed, connected or deployed. When we finish there is no FuseLoom system running anywhere in your business.
- Not a deck delivered from a distance. The thinking happens in sessions with your people; the document is the record of what was decided, not a substitute for deciding.
- Not a technology shopping list. Tools get named where naming one helps, with the reason attached — and the reason is never that we have a stake in it.
- Not a licence to build. The map recommends and you decide. Building is a separate conversation, and you are welcome to have it with someone else.
When we'd tell you not to take it on
- You already know the one thing that hurts. If your team can name the task without a workshop, you don't need a strategy engagement. Bring the task to a discovery call — the map that comes with the build will cover it.
- The real blocker is a decision nobody has made. A roadmap does not settle an argument about direction — it only makes the argument better documented.
- Nobody on your side can own it. A plan without an owner is a document. That person does not have to be technical, but they do have to exist.
- The honest answer is "not yet". When the right next step is to run the process by hand a while longer, you hear that on the call, before you have committed to anything.
Four questions, answered before you ask them.
Set down in writing rather than saved for the meeting, so nothing important depends on how well a conversation goes — and so you can forward the answer to whoever else has to agree.
Q01 If it isn't software, what am I actually getting?
Judgement, and the hours it takes to earn it: interviews with your people, a working session where every idea gets scored, a drafted plan, and a session where you take it apart. The people doing all of that are the people who would build whatever you decide to build — which is why the recommendations are things that can actually be built.
What you keep is the paperwork: the opportunity map, the phased plan, the skip list, the readiness check. It stays useful whoever ends up doing the work.
Q02 What would you be committing to?
Nothing about the shape of this engagement is settled before anyone has looked at your work, because a scope named that early is a guess. What the work involves follows scope: how many people are interviewed, how many processes are mapped, and how much of your data has to be checked before an answer means anything.
After the 30-minute call you get a short written proposal with the scope on it, agreed before anything starts — and neither the call nor the proposal commits you to anything. You can stop at the end of any stage, keep what has been made, and carry no commitment for the stages you never approved.
Q03 Do you need access to our data to do this?
Only where the readiness check has to look at real records — and never before it is written down. The mapping itself runs on conversations and on the shape of your systems rather than on their contents. Where real records do have to be opened, we agree in writing what may be looked at, what may be kept, and what may leave your organisation — and you sign that before anything is opened.
An NDA comes back signed before you send us a document or a login. Your data is never used to train AI models, and that sits in the contract rather than on a webpage.
Q04 Will the roadmap just recommend more work from you?
It tells you what to do first. Some of that will be work FuseLoom could do, and we will say so plainly rather than pretend to be neutral. The parts that prove the map was not written as a pitch are the skip list and the items marked "keep doing this by hand for now" — neither of them is there to win work, and neither comes out of the document.
You can hand the whole plan to another firm and it will still make sense. It is written to be read by whoever ends up doing the building.
Bring the ideas. Leave with the order.
Half an hour with us. We look at what you are being pushed to do with AI, say plainly which part of it is real — and tell you if you don't need a roadmap at all.
Ask us to call you backwith FuseLoom · [email protected] · the call, and the written proposal that follows it, commit you to nothing