How to efficiently push business teams to use AI

This one is for AI leads who own adoption: getting business stakeholders, the people in finance, sales, operations, and marketing, to use AI in their daily work.

These are my thoughts after two years of leading AI adoption at Danish Endurance, a fast-growing company of 100-plus people. It's what has worked for me there, not a universal playbook.

Most of us came to the role from code, and it's easy to measure our impact by what we build ourselves. But a company doesn't get faster because the AI lead is fast. It gets faster when the business is.

The catch: AI leads own a lot of things now, and adoption is only one of them. So it has to run efficiently: as many business people as possible using AI well, without it eating all your time or making a mess.

Here's how to push business teams to use AI efficiently.

Start with the obvious: LLMs in Excel, Word, Outlook, and PowerPoint, the win we overlook.
Steer from the org level: org-wide instructions and skills, managed in GitHub.
Run a champions program with accountability: a group that shares, reports back, and brings ideas.
Measure changed workflows, not seats: department heads own the outcome.
Put a price on the hours saved: leadership needs to see the impact in money.
Manage the excitement: enthusiasm without coordination becomes silos and double work.

Start with the obvious

If you take one thing from this post, take this one. It's the easiest thing to implement and to push.

We're biased. For someone who used to code, LLMs in office tools feel too obvious to be a strategy. But that's where business stakeholders work. The harsh reality is that they spend an enormous amount of time in Excel, Word, and Outlook, and that time slows the whole organization down. At this point there's no reason for anyone to work in a spreadsheet without an LLM next to it.

Push it the way you'd push any tool rollout:

Make it available: the add-ins installed and licensed, for everyone, not only the people who ask.
Show it on real work: a team's own spreadsheet, a reply to an actual email, last month's deck.
Make it consistent: shared skills for PowerPoint and Excel (saved instructions and templates the LLM loads for the task), so every deck and every workbook comes out in the same layout.

Steer from the org level

The main LLM tools now let you set instructions and skills for the whole organization. It's an incredibly powerful lever: everyone starts from the defaults you set, so you steer the company toward efficient use of AI without teaching it one person at a time.

Org-level instructions: the context every conversation in the company starts from.
Skills for recurring work: a PowerPoint skill that builds every deck in the company's style.
Branding: image assets that ship with the skills, so what the LLM produces looks like the brand.

Manage all of it through GitHub. The skills and assets live in one repository, versioned and reviewed like any other code, and a change to the deck style reaches the whole company at once. You even control whether each skill is installed for everyone by default or left for people to switch on.

That's where the extra value sits. You write a skill once, and a sales manager's deck or a controller's workbook looks like the company's, not like whatever the model guessed.

Run a champions program with accountability

Everyone knows the idea: an AI champions program, or an AI tribe. A group of people from inside the business teams whose job is to help their colleagues use AI more efficiently. A champion in finance reaches finance in a way the AI lead can't.

The whole goal is to make AI click in people's heads, so they get interested in the topic itself. Once it clicks, people are very capable of pursuing their own path with it. You don't have to hand them every use case; you have to get them to the point where they go looking.

Most of these quietly stall. The big mistake is to share something with the champions and then wait for them to use it, or to pass it on. To make it work, you need to manage it:

A recurring meeting: the same people, every month.
Accountability for sharing: each champion is responsible for bringing what you shared back to their team.
A system to track it: what you shared, with whom, and whether it reached their team. Ask about it in the next meeting.
New ideas, not just new tools: champions are also accountable for bringing ideas from their teams back to the group, and that gets recognized.

In practice, it looks like this. Meet each champion once a month, one to one. If the company is bigger, meet small groups instead. The smaller the group, the better, as long as you can still manage it: in a small room you hear each team's real problems, not the polite version.

In each meeting, you own two things:

Bring something relevant to them: a new skill, an artifact, or any new feature that fits their work. Not a tour of everything that shipped this month. Relevance is what makes it click.
Understand who you're talking to: how technical this stakeholder is, and whether their department actually benefits from a lot of AI. A finance team living in spreadsheets and a team that rarely touches a computer need very different things from you.

Pay attention to both, every time. Then track what happened. A shared tracker is enough (example rows):

Date	Shared	Champion	Shared with team?	New idea brought
March	Excel add-in for forecasting	Sales	Yes	Weekly report from CRM export
March	Excel add-in for forecasting	Finance	Not yet
April	PowerPoint skill	Ops	Yes	Supplier review deck

Then create urgency. A few months of someone not using AI in Excel costs a lot of money and slows the organization down. That's the case to make in every meeting.

It also matters for you. A working champions group means the whole company isn't relying on the AI lead to move.

Measure changed workflows, not seats

Most AI leads report licenses and active users, because the tools hand you those numbers for free. They look good while nothing changes. Someone who opens the Excel add-in twice a month counts as active, and so does a team that rebuilt its weekly report around AI.

Measure workflows instead: "the weekly sales report used to take half a day, now it takes an hour." A handful of those says more than any usage dashboard, and they're the evidence behind the urgency you're asking for.

Then give the outcome an owner. Ask each department head for one or two workflows to change, with a before and an after. Champions spread the tools; managers own the result. That closes the gap most champions programs have: the champion is accountable, but their manager isn't.

Add a column to the champions tracker for it, and bring every changed workflow back to the next meeting. A workflow that changed in one department is the best pitch to the next one.

Put a price on the hours saved

Leadership has a hard time seeing this work. Champions meetings and workshops don't show up in the reports they read, and depending on your leadership, the whole program can look like time you could have spent better. You need financial ground under it.

So quantify it. Run surveys with the business teams and ask how many hours AI saves them per week or per month. The changed workflows tell you where to ask.

The rule of thumb is simple: estimate the hourly salary per employee, and multiply it by the hours they save. Add it up across the teams, and you have a monthly figure in money.

It won't be precise, and it doesn't need to be. It turns "people use AI more" into a number leadership already knows how to weigh.

Manage the excitement

Adoption has a messy phase. Once business users get excited about AI, they start building: a side project here, an automation there. A lot of it is siloed, some of it is the same thing built twice, and little of it adds up to capacity the organization can actually use. Left alone, that becomes a gigantic inefficiency.

You can't manage it with one announcement. As the AI lead, you have to show your face and repeat the message, again and again:

In emails: what already exists, so nobody rebuilds it.
In presentations: which projects the company backs, and why.
Through the champions: they hear about the side projects first, so they're the ones who can point people to what's already there.

Let people experiment. Just make sure the experiments find each other.

Everyone else's speed

None of this is technical, and that's why AI leads skip it. It's meetings, trackers, surveys, emails, and saying the same thing for the fifth time.

But it's where the impact is. Your own projects make you faster. Office tools, a champions group that delivers, and less duplicated work make the business faster, and that's the job.

Pick one team that still works in spreadsheets without an LLM, and start there this week.