Contents · 6 min read
01 / 05 · The gate
Why should the management team go first?
In short
Start with the management team. It must use AI in its own work before choosing where the company starts. That gives the choice a view of the whole business.
A pharmaceutical affiliate has five commercial functions. Key launch signals still move through a quarterly meeting. Its management team has used AI on real work. That shared view helps leaders look past a fast local gain and choose work that can speed the whole launch.
The affiliate’s management team has now used AI in its own work. It has used AI to prepare and analyze real work. The team now has to choose which function goes first.
That is why an AI-augmented management team comes before a function plan. Its job is to set company direction and resolve shared choices.
For the affiliate, a rank by likely return seems sound. Each head can show where work is slow. The list can reveal useful work.
Yet the scores start inside each function. They do not show the main change across all five. The choice now rests on what changed at the core.
02 / 05 · Core change
What changed at the core?
In short
The main change is faster action across the five functions. A rank of separate teams would miss that. The company must first find the delay that no function owns alone.
For the affiliate, market access may spot a formulary change. That fact should shape medical priorities and field messages within days. Today, it may wait for the next cross-team meeting.
Each function can be strong on its own. The slow part is how facts move between them. The pharma CEO briefing shows why five strong functions can still lose launch speed.
A sales pilot could save time. A medical pilot could do the same. That is useful work. But neither one has to fix the handoff between them.
For the affiliate, local gains could leave the launch delay in place. The first build must match the shared change.
03 / 05 · First build
Start with one layer around one launch
In short
Pick one launch. Build one shared intelligence layer around it. Let that layer watch for new facts and send them to the right teams.
For the affiliate, the layer links all five functions around one live launch. A formulary change can reach medical affairs and the field team while it still matters. A launch signal can prompt a shared response.
The pharma operations briefing makes the choice plain. One layer around one launch comes before five function pilots. It changes how the work joins up.
The five pilots are still a fair contrast. Each may cut local work. Each may give its function a useful tool. Yet all five can leave the old handoffs in place.
For the affiliate, people still own each call. The shared layer prepares work. It keeps memory and follows through. That same split runs through an AI-augmented company.
Once the shared layer works, the team can set the order for each function.
04 / 05 · The order
Which function comes next?
In short
Rank each function by its value to the core change. Then check if its team is ready. Check shared needs before you set the order.
Strategic value
Ask how much this function can speed the shared launch response. Count the effect on the company. Do not count only hours saved inside the team.
Readiness
Ask if the team can start with material it can share now. A safe test may move sooner than work that needs live system access.
Shared dependencies
Ask what must be in place first. If several teams need the same facts, move that shared work up the list.
Revenue may show value. Capacity may too. Risk can push a team back. Learning can move shared work forward. No one measure should set the order for every firm.
For the affiliate, a strong local case can move down the list. That happens when it relies on shared work that does not yet exist. A ready team can move up when its work also helps the core change.
This is the point of starting with what AI changes in the business. Good AI consulting maps the highest-value places to move from that change. It does not begin with a list of tools.
The first choice still needs proof before the next one gets funds.
05 / 05 · The proof
Make the first step earn the next
In short
Fund the shared layer as one clear step. Set the proof before work starts. Add the next function only when the result supports it.
For the affiliate, proof should show whether launch facts move faster. It should show whether the right teams act sooner. The team can set its own bar before work starts.
Four steps keep that call clear:
Name the result
The CEO and management team state the change they want. Here, it is faster action across the five functions.
Set the proof
The team agrees what it will track. Use the time from a new signal to a changed launch action.
Run one clear step
The work owner builds around one launch. The team records what changed.
Make the funding call
The management team reviews the proof. It then backs the next function. If proof is weak, it changes the work or stops.
There is no fixed financial bar in the source work. There is also no set count of functions to run at once. Start with the shared layer. Add work in parallel once leaders can see the impact.
For the affiliate, this keeps a weak local win from setting the whole roadmap. It also lets a strong result earn more funds. Each part stays self-contained, which is how the path from proof to wider use works.
The next view is how that choice works across the whole company.
Set the sequence
The management team’s next task is to choose the shared business change and set a sequence that evidence can advance. See The Opportunity helps the team build the fluency to make that choice through hands-on use of AI in its own work. Across three workshops, GRAIL then turns that shared view into a ranked roadmap, so the first move reflects company value and dependencies rather than the loudest local case.
Three to four weeks, three workshops with your management team: hands-on use of AI, custom AI agents for each leader, and a ranked roadmap of opportunities. Fixed fee, no access to your live systems.
Questions CEOs ask about this
Which business function should our company transform with AI first?
Start with the management team, not a business function chosen in isolation. The CEO must decide whether shared work comes first, because a local win can leave the company-wide delay intact. Rank later functions by their value to that change. Then check readiness and shared needs. Proof should earn the next step.
Which goal should set the order?
Start with the core business change. Revenue or capacity may show its value. Risk can limit the scope, while learning tells you what the first step must prove.
How do shared needs change the order?
If several functions need the same facts or signal route, build that common part first. A local tool should wait when it depends on shared work that is still missing.
When should the management team go before the functions?
It should go first when leaders have not used AI enough to see what changes. The team needs that shared view before it can set a sound order.
How many functions can we redesign at once?
Do not force a fixed number from this test. Start with one clear build, then add work in parallel when leaders have seen the impact and can support the next wave.
What evidence is enough to fund the next function?
Set the bar before the first step starts. For the affiliate, proof means signals reach the right teams faster and change launch work sooner. There is no standard financial threshold that fits every company.
Founder, GRAIL
Johan has spent fifteen years in management consulting and executive leadership, including as Chief Product Officer of a SaaS company. He founded GRAIL to help European leadership teams work out what AI changes at the core of their business. He is the author of The Execution Revolution and the host of the ThinkRoom podcast.
Related reading
Sources
- What does an AI-augmented management team look like?: why company choices start with the team that sets direction. (accessed October 8, 2026)
- AI Consulting for the Nordics: leadership fluency first, then work where the return is highest. (accessed October 8, 2026)
- The Patent Clock Is Running: the five-function gap and the slow quarterly cycle. (accessed October 8, 2026)
- What’s Changing in How You Launch: one layer around one launch instead of five pilots. (accessed October 8, 2026)
- What Is AI Consulting?: start with the core business change and map where to move first. (accessed October 8, 2026)
- What Does an AI-Augmented Company Look Like?: what the shared layer takes on and what stays with people. (accessed October 8, 2026)
- What we do: self-contained steps that move from proof to wider use. (accessed October 8, 2026)
Next scheduled review: February 15, 2027