The Embedded Unit
We build it to become unnecessary.
Six to eighteen months · Priced against the decision
You have been handed a mandate you cannot hire your way out of.
The capability does not exist in the business. The people who could build it are not available, the market rate for hiring them is not in the budget, and the training programme you ran last year changed nothing that survived contact with the day job.
Roughly 95% of enterprise AI pilots produce no measurable impact on profit.
Not because the technology fails. Because capability lives in individuals rather than in the institution. Individuals leave, get promoted, or go back to what they were doing before.
Training does not fix that. A department does not fix it either. It just gives the organisation somewhere to send difficult work.
What happens
Design the unit
Its remit, its operating model, where it sits, what it is allowed to do, and what it is allowed to refuse.
Name it and brand it
A unit people can join, refer to and be proud of is a different object from a workstream. This is how a way of working acquires an identity that outlives a budget cycle.
Teach through live work
Not a course. People learn by doing real work with real deadlines and real clients on the other end. Nobody is taught a method in a room and sent back to their desk to apply it.
Model its ROI
A unit that cannot demonstrate its own return does not survive the next budget round, however good it is. A simple model is better than no model.
Hand it over
To named internal leadership, with the capability already proven on work that shipped.
What success looks like
The unit stops existing.
An embedded unit still standing five years on has usually become another department. A place difficult work gets sent, rather than a change in how work gets done. That is the failure mode, and it is comfortable enough that most people never notice it happening.
Success is the unit being absorbed into the way the organisation works, and no longer being needed as a place.
We build it to become unnecessary.
The proof
Makerspace at Ogilvy & Mather
Ogilvy had a structural problem dressed as a cultural one. Ideas were discussed, planned and researched for months before anything existed that a client could look at. As the group's chief creative officer put it publicly at the time: "The amount of chat that goes on in an agency is painful. It can take up to a year to make a TV commercial."
Strategists were rewarded for thinking. Making happened somewhere else, later, by someone else.
So we built a unit. Created with Ogilvy's EMEA C-suite, with Gary Bonilla as strategic maker brought in to lead it. Makerspace had a name, a logo, a manual, a curriculum, a home on the first floor by the entrance, and a sign that read proudly making pilots in public.
One flat fee of £10,000. Under thirty days. Every time.
Teams brought their solutions to client briefs. We built a pilot, tested it, iterated, and handed back a working demo with the data to support the sale. Rather than asking for new budget, Makerspace redirected existing budget: instead of spending £20,000–£30,000 planning and researching an idea, take £10,000 of that and make something.
Five makers on staff. Fifteen strategists trained through it. Three for every maker, because the unit existed to change how people worked, not to absorb the work.
And then Makerspace stopped being a place. It was embedded into the wider organisation as a new way of working.
The unit was the intervention. The client work was the evidence it took. Its disappearance was the result.
The AI version
The same play, for the problem most organisations now have
AI capability sitting in individuals. No shared standards, no institutional guidance, no line of sight between what the team does internally and what the organisation produces externally. Pilots that work and never scale.
An embedded unit installs the capability where it survives. In the institution, in the way of working, in people who learned by shipping rather than attending.
We have run this play twice, in hostile conditions, and it worked both times.
What you get
- Unit design and operating model
- Identity: name, positioning, and a way of working people can articulate
- Capability built through live engagements, with delivered output as the evidence
- An ROI model the unit can defend itself with
- Handover to named leadership, with a continuation plan
What it costs
Priced against the decision, not the days. Scoped to the size of the capability gap and what closing it is worth. Fixed fee once scoped, invoiced across the engagement.
What it is not
- Not a training programme
- Not an innovation lab
- Not an internal agency
- Not a centre of excellence that produces a framework nobody uses
- Not permanent. If it is still there in five years, it failed
This is the deepest engagement we run and it rarely starts cold. Most Embedded Units begin as a Pilot Case or a Transformation Pilot that finds the constraint is capability rather than strategy.
If you already know that is your problem, the call is short.