You've set the company’s AI strategy and bought the team licenses, but whether any of it changes how work gets done now depends on your managers – because they're the ones who translate strategy into day-to-day behavior. 

Unsurprisingly, this is the layer where most AI transformations stall.

Articulate, which makes software for building workplace training, is an interesting case study because it has no head of AI. Each executive owns AI in their own function, and managers carry the change. 

We recently sat down with Monika Saha, Articulate's Chief Commercial Officer, to talk about how she sets her go-to-market managers up to lead it. Here's what we'd take from her approach.

1. Help managers narrow their scope

Any team can list dozens of processes AI might automate. Monika avoids the overwhelm of use case selection by asking each of her leaders for exactly two processes (and no more) that are painful or holding the team back, to force them to focus on where the opportunity is highest. 

And a written brief is not good enough – each VP records a short video of how the process works today. If they're too far from the work to show it, they connect with the people who do it. She uses these videos to estimate what fixing it would be worth to the company, usually in hours or dollars. Now they have a baseline for proving success. 

For example, buying new software at Articulate used to mean rounds of emails with IT, compliance and legal. An AI assistant that knows the purchasing rules now answers those questions directly and saves up to 20 hours a week. 

2. Give managers more ownership

Monika says quality standards and judgment should sit with each function, even in companies that have a head of AI. Owning those standards is what makes managers the people leading the change, rather than recipients of tools someone else built.

Articulate's legal team is a good example: When the company added AI features to its product, its customer contracts needed new clauses, and customers started asking sales reps what they meant. Legal built a Claude skill that reps can use to ask about any clause, and legal decided what Claude can and can't go into detail on. The experts control the quality, and sales gets answers without waiting.

Make it clear that functional leaders continue to be the domain experts, and give them the authority to set the standards for what good looks like.

3. Make all AI projects public in your chat 

Articulate pushed the company to run AI projects in public Slack channels, where teams post progress as they go. Anyone can ask the company's AI specialists in IT for help in this open channel, so the answers are visible to everyone.

For managers, this is how you keep oversight without adding status meetings. You can see when a team is building something heavier or more expensive than the problem needs and redirect it early. 

Wins get recognized while they're fresh and other teams get to learn from exposure rather than siloed experience. More cost effective for you, more empowering (and less scary) for them.

4. Expect conversation over perfection

Managers are learning AI at the same time as their teams, and whatever anyone knows today will be outdated in a month. So don’t expect managers to become experts, but do expect them to show their teams what they've tried and where they're stuck, to allow everyone to learn together rather than fail privately. 

Monika rejects the idea that AI enablement is a second job for managers who are already overloaded. Coaching has always been part of managing – that’s all this is.

5. Prepare them for the people side of it

Your managers are not immune to the anxiety of AI-driven changes. They are being asked to lead through a moment they’re not experts in. Your job is to equip them to show up consistently in a way that motivates their teams.

Set the expectation of repetition. One demo of a hot new AI use case, or a small AI win, rarely convinces people to stop working the way they’re used to. Managers need to establish a regular cadence of AI demos and check-ins. 

For example, if two people on a team of eight adopt a new workflow, their wins should be highlighted, and the other six should each be asked what's getting in the way. 

What to do if you only do one thing

Put the two-problem exercise on your next leadership agenda.

This is the step that makes the rest possible. In the process of bringing you two recorded processes with a cost attached, they uncovered the biggest gaps in their teams, the value they could be creating, and something concrete to build.