Your Company Is Going AI-Forward. Here’s What Leaders Should Do First

Asking for a Friend…

“Our CEO just made a big announcement that we’re going to be ‘AI-forward’ and wants every department looking for ways to transition work to AI agents over the next six months. Now what?

I'm hearing versions of this question more and more. You may be wrestling with it too.

Your CEO has established a direction. Now comes the leadership work of translating “AI-forward” into better work, stronger capabilities, and thoughtful experiments your team can learn from.

Summary: Your First Moves as an AI-Forward Leader

Start by using AI yourself so you understand both its possibilities and its quirks. Then get curious about what your employees are already doing, where work is unnecessarily difficult, and what they think AI could improve.

From there, identify a few meaningful quick wins, get very clear about the human and AI roles, run bounded experiments, and create a rhythm for learning from what happens.

Most important, involve your people throughout the process.

AI-forward leadership will require technology choices. It will also require the same leadership fundamentals that help teams navigate any significant change: Connection, Clarity, Curiosity, and Commitment.

1. Go First: Build Your Own AI Fluency

If you want your team to experiment intelligently with AI, they need to see you learning too.

Pick work you know well enough to recognize a good result from a lousy one.

Use AI to prepare for a meeting. Analyze a set of information. Ask it to challenge your thinking. Create an agent to monitor something repetitive. Experiment with a task you’ve done enough times to understand where AI genuinely improves the work and where it gets weird.

Pay attention to your own reactions.

Where did you trust the output too quickly?
What required more context than you expected?
Which task became dramatically easier?
Where did you spend more time fixing the AI than you would have spent doing the work yourself?
What surprised you?

Then talk about what you’re learning.

You'll make people feel safe to try themselves when you model the way.

“I tried this. Here’s where it helped. Here’s where it fell apart. Here’s what I’m trying next.”

In our book Courageous Cultures, we talk about how culture begins with leaders modeling what they want to see. If experimentation, learning, and speaking up matter in your AI-forward culture, be one of the first people doing those things.

2. Go on an AI Curiosity Tour

Before assigning everyone an AI initiative, find out what’s already happening.

There’s a good chance people on your team are further along than you realize.

Someone is using AI every day and has figured out three tricks nobody else knows.
Another guy tried it twice, got terrible results, and decided the whole thing was ridiculous (and is telling everyone his woes)
Another employee has quietly built a workflow that saves two hours every Friday, but doesn't want to tell anyone for fear of looking lazy.
An contact center agent else sees a use case that could solve a customer frustration you’ve been discussing for a year, but no one has asked her what she thinks

And there may be people experimenting with company information in ways you’ll want to understand sooner rather than later.

Go ask.

A good AI Curiosity Tour is about showing up genuinely curious:

Note: If you haven't seen my Asking for a Friend Video on Curiosity tours filmed in New Zealand (it's fun and filled with practical advice).

“Where are you already using AI in your work?”
“Show me something that’s working surprisingly well.”
“Where have you tried AI and found it wasn’t helpful?”
“What part of your job would you happily never do again?”
“Where are we making work harder than it needs to be?”
If you had a capable AI agent working beside you, what would you want it to take on?”
“Where would you be uncomfortable having AI involved?”
“What do you know about this work that an AI system is likely to miss?”

And one of our favorite Courageous Questions:

“What's one thing people saying about AI to one another that they aren’t saying to leadership?”

Listen for both opportunity and concern.

In our Courageous Cultures research, one of the most common reasons people didn’t share ideas was beautifully simple: no one asked.

An open-door policy won’t fix that. Go find the thinking.

3. Look for Work Worth Improving

Once you’ve listened, resist the urge to create a list titled “Jobs We Can Give to AI.”

Look at the work instead.

A job contains all kinds of activities—some repetitive, some relational, some analytical, some emotionally complex, some requiring years of judgment.

You might discover that AI can take eighty percent of one process while another role benefits most from AI helping a human think better.

We’ve found it useful to consider four ways work can be designed:

Human-led: A person owns and performs the work.

Human-led + AI-assisted: A person directs the work while AI researches, analyzes, drafts, simulates, or recommends.

Agent-led + human-reviewed: AI performs much of the workflow while people review defined outputs, decisions, or exceptions.

Agent-executed within guardrails: AI completes a bounded workflow and escalates when defined conditions occur.

The goal isn’t to push as much work as possible toward the bottom of that list.

You’re looking for the right relationship between human judgment and AI capability.

That brings you back to a practical question:

Where could we make the work meaningfully better?

Better for the customer and employees.
Faster where speed matters.
More thoughtful where judgment matters.
More consistent where inconsistency causes problems.

And perhaps a little less soul-sucking where people are spending four hours every Thursday copying information from one system into another.

4. Pick a Few Quick Wins Worth Learning From

You’ve probably got more possibilities than capacity.

Great. Pick a few that will teach you something useful.

This is a good place to use our I.D.E.A. model from Courageous Cultures.

Ask whether the opportunity is Interesting: What meaningful problem does this solve? How would it improve the customer or employee experience, efficiency, quality, or results?

Then consider whether it’s Doable: Could we realistically test this with the systems, information, permissions, people, and time we have?

Think about Engaging: Who needs to be involved? Who understands the work? Where might you meet reasonable resistance that could help you improve the idea?

Finally, identify the Actions: What are the smallest meaningful steps we can take to test this?

A good first experiment usually has a visible benefit, manageable risk, people who care about making it work, and enough complexity that you’ll learn something.

You’re trying to build capability along with results.

5. Get Clear Before You Give an Agent More Autonomy

This is where your enthusiasm needs a good partnership with Clarity.

Before an agent starts doing meaningful work, decide what it can do.

Can it read information?
Recommend an action?
Draft something?
Send it?
Change a customer record?
Issue a refund?
Approve an expense?
Communicate directly with an employee?

The verbs matter.

So do the handoffs.

When does a person review the work?
What causes an escalation?
Which decisions remain human?
Who can change the agent’s permissions?
Who can stop the workflow?

Match your guardrails to the level of risk.

An agent summarizing meeting notes carries different consequences than one communicating with an upset customer or making a financial commitment.

One question can help your team calibrate the level of autonomy:

How consequential is it if the AI gets this wrong, and how quickly will we know?

Your answer tells you a lot about how much oversight you need.

Article content

Important questions to ask when leading AI change.

6. Create Small Experiments With Big Learning Loops

Decide what you’re trying to learn.

Maybe you want to know whether an agent can handle a certain category of requests accurately.

Perhaps you’re testing whether AI assistance improves the quality of managers’ coaching preparation or performance appraisal feedback.

You might be exploring whether an automated workflow saves time without making the customer experience feel robotic.

Choose a few meaningful measures.

Then schedule the conversation where you’ll review what happened.

Ask:

What worked better than we expected?
Where did humans add the most value?
Where did people override the AI?
What did customers notice?
What frustrated the team?
Which assumption turned out to be wrong?
What should we change before the next round?

When employees take the risk of experimenting, sharing concerns, or pointing out flaws, show them what happens with what you learned.

That response determines whether they’ll keep talking.

7. Talk About the Learning, Including the Messy Parts

Your team is paying attention to the stories you tell about AI.

If...

every story is about efficiency, people will make assumptions about what you value.
experiments have to become a success story, people will quickly learn to hide the failures.
the person who catches an AI mistake gets blamed for slowing things down, good luck getting people to raise their hands next time.

Narrate a different culture.

Talk about the pilot that saved eight hours.
Also talk about the experiment that failed because the workflow made no sense.
Celebrate the employee who discovered a better use case.
Thank the person who said, “I think the AI is wrong,” and turned out to be right.
Share the example where a customer clearly needed a human being.
Discuss the skill your team realized they need to strengthen because AI is changing the work around it.

People pay close attention to what gets celebrated, what gets questioned, and what happens when someone speaks up.

That’s how “AI-forward” becomes culture instead of a slogan.

Use the 4Cs as a Leadership Check

As your experiments grow, our performance framework (Connection, Clarity, Curiosity, and Commitment) gives you a useful way to notice what might be missing.

Connection: Are the people closest to the work involved? Do they understand why the work is changing? Are you hearing their ideas and concerns?

Clarity: Does everyone understand what belongs with humans, what belongs with AI, where the boundaries are, and when a person needs to step in?

Curiosity: Are you experimenting and learning? Are people questioning AI recommendations as readily as they question human ones? What are you discovering that changes your thinking?

Commitment: Who owns the outcome? How are you monitoring performance? What happens when something goes wrong? What are you doing with what you learn?

You can use those questions whether your department is running its first small experiment or coordinating a growing collection of AI agents.

What I’d Do in Your First 90 Days

If we were talking over coffee, I’d keep your first few months pretty simple.

During the first couple of weeks, use the technology yourself and listen aggressively. Learn enough to ask better questions. Take your curiosity tour. Find out what your people already know.

Over the next month, identify two or three meaningful opportunities and design a few bounded experiments. Bring the people who know the work into the design. Get clear about decision rights, human checkpoints, and what you’re trying to learn.

Then run the experiments, review what happens, and make the learning visible. Expand what works. Change what almost works. Stop what isn’t creating value.

By the time you reach ninety days, you won’t have your department’s entire AI future mapped out.

You’ll have something more useful: people building fluency, a few real examples, better questions, early guardrails, and a clearer understanding of where human + AI collaboration can improve the work.

That gives you a much stronger foundation for the next three months.

One More Courageous Question

Your CEO has told you the organization is going AI-forward.

Before you start assigning pilots, ask your team:

“If we do this really well, what could become possible for our customers, our team, and the work itself?”

Then listen.

There’s a decent chance your best first move is already sitting in the room.