More capacity only helps if someone decides where it goes.
AI is substantially changing how teams work every day. And there’s plenty of pressure surrounding that change. Another tool to try. Another task that could be automated. Another way to do more work, faster.
The appeal is obvious. But there’s a trap here. Faster work doesn’t automatically mean better performance. More often, it means AI solves for speed. Then it exposes all the places where speed was never the problem.
Teams get more done without improving outcomes because knowing what’s worth doing was always the harder part. In my work leading Duetto’s product team through AI adoption, it’s why I’ve learned to start with a different question.
Not “can AI do this?” But “what gets better if it does?”
If the answer stops at ‘more time for more output’, we haven’t gone far enough. Leaders still have to decide what to do with the capacity AI creates and where it can have the greatest impact.
AI can help your revenue team analyze pickup, competitor movements, and demand signals in a fraction of the time. Great. What happens next?
The decision may still depend on other teams. Sales and marketing may be working from a different view of demand. Finance may need to understand the impact. And across a portfolio, someone still has to turn that insight into action across multiple properties.
Speed up one part of the process and sometimes you just reach the next bottleneck sooner.
The same problem shows up when teams approach AI in isolation. One team spends time building a workflow another team has already figured out. Different tools produce different answers to the same question.
Without shared thinking around tools, data, and where AI actually belongs, you can end up with five teams getting faster in five different directions.
The instinct is to fix that with more people. A task force, a working group, a standing meeting with half the company on the invite. That rarely makes it faster. What works better is naming one person to own the pattern. They define how the work gets done, which tools and data it uses and where a person steps in, and they write it down. Everyone else starts from that recipe and adapts it. One owner for the recipe, and everyone else gets to cook.
Try this week
Pick one workflow your teams keep reinventing and name who owns it.
There’s another side to consider with all the time AI promises to free up. Efficiency has a habit of becoming expectation.
What do I mean by this? A report that once took an hour now takes 10 minutes, and suddenly there’s time for another. Your team can analyze more scenarios, so now there are more scenarios to analyze.
The plate never gets lighter. We just find more to put on it.
Over time, that crowds out the deeper thinking all that extra capacity could’ve made room for. And it can take a toll on people too. In 2024 Quantum Workplace research, 45% of frequent AI users reported high burnout compared with 35% of people who never use AI.
45%
of frequent AI users reported high burnout
35%
of people who never use AI
It’s the job of leaders to protect the capacity AI creates. An hour saved on reporting doesn’t have to become an hour of more reporting. It can become time to challenge the strategy, investigate an unexpected shift in demand, or bring sales and marketing into the conversation.
That’s a better measure of AI’s value. Not just the hours it gives back, but whether your team can put them toward work that improves performance.
Adopting more AI tools isn’t the same as getting more value from it.
There’s a temptation to race toward the most advanced use case available. But the closer AI gets to taking action on behalf of your team, the more deliberate you need to be about how you get there.
Our CTO, Bob Matsuoka, outlines a useful progression for hotels.
Explore your data.
Start by asking questions of your hotel data and validating the answers against what you know.
Generate insights.
Use AI to surface patterns, anomalies, and opportunities that deserve a closer look.
Get recommendations.
Ask AI to suggest what you should do next based on what it sees in the data.
Automate actions.
Once you trust the inputs and outputs, allow AI to take defined actions and trigger workflows on your behalf.
Each step should earn the next one. Think of them as checkpoints. If AI surfaces an unexpected shift in demand, can your revenue team validate it? If it recommends a pricing action, does that recommendation hold up against what the team knows about the market?
Those answers tell you whether you’re ready to give AI more responsibility. Build confidence at each level before you automate the next one.
The further you move through the four steps above, the more important it is to know what your AI is working with.
Automate a messy process and you get a messy process that moves faster. Give AI inconsistent definitions, incomplete data, or unclear rules, and those problems become part of what gets scaled.
The stakes rise as AI moves from surfacing information to recommending and taking action. The more responsibility you give it, the fewer chances there are for someone to catch a problem before it influences what happens next.
That’s where leaders need to set the guardrails. Is the data reliable? Are teams working from the same definitions? Who owns the decisions AI informs? Where can AI act and where does a person still need to step in? Good governance has to account for both the technology and the information feeding it.
And when an approach does work, standardization gives other teams and properties a proven starting point, rather than asking each of them to solve the same problem again.
AI can accelerate an extraordinary amount of work. But it can’t take two of the hardest responsibilities off our hands.
Knowing which problems are worth solving.
Knowing how to lead people through the noise.
Those are still leadership calls. And they matter even more when our teams are navigating changing tools, workflows, and expectations for what they can accomplish.
That’s why in my own work I use a pretty simple test for AI. Does this give me more room to lead?
I’ve built a system in Claude for that purpose. Before a one-to-one, I use it to revisit past conversations, goals, and reviews and quickly reconnect the dots on what needs attention. Because what I really want back isn’t just time. It’s more attention for the part of the job that needs me.
At an organizational level, the principle is the same. More capacity only helps if we’re clear about where to point it.
This is where a performance engineering mindset helps. Profit gives everyone a shared North Star. Instead of asking what AI can do for revenue, sales, marketing, or finance in isolation, leaders can ask where it will help the hotel make better decisions and improve performance as a whole.
You might also like: The what and why of performance engineering.
AI can give your team hours back. It can’t decide where those hours go. That’s the manager’s job.
~70%
of the variance in team engagement is accounted for by managers
More than a decade of Gallup research has found that managers account for roughly 70% of the variance in team engagement. Every problem in this article lands on a manager’s desk. The saved hour that turns into more reporting. The recommendation nobody knows whether to trust. The five workflows that should be one. No tool fixes those. A manager who notices, asks the right question and makes the call does.
Here’s where I’d start and you can do all three this week.
Ask in your next one-to-one
what AI gave back this month and where it went.
Protect one block of that time
for work only your team can do, like challenging the strategy or chasing an odd shift in demand.
Name one owner.
Pick one person to own how your team uses AI on a recurring workflow and ask them to write it down.
So is AI making teams better? It can when someone decides where the extra capacity goes. That someone is the manager. That’s good news because it’s a decision any of us can make this week.
AI creates the room. What happens in it is up to us.
You might also like: How hotels can conquer complexity with AI.