AI is everywhere. It drafts emails, summarizes meetings, analyzes data, and writes a first pass of project briefs. On an individual level, it’s working. In our research at Atlassian, we found that knowledge workers say AI is helping them save 76 minutes per day. That’s the shallow end – useful, energizing, and a good way to build momentum.

But here’s the part we don’t talk about enough: most of that speed isn’t yet showing up in team-level outcomes. Why is that?

According to Atlassian’s AI Collaboration Index, only 4% of executives say AI is helping their teams solve previously unsolvable problems. And 37% say AI has caused teams to waste time or head in the wrong direction. Those aren’t technology issues alone – they’re culture and coordination issues.

I’ve been digging into this gap and rewatched sessions from Teamwork in an AI era, featuring Atlassian’s Teamwork Lab, Forrester researchers, and Adam Grant. A pattern emerged: leaders frame AI as an individual productivity tool, while the real shift lies in how teams coordinate, share context, and build trust together.

The good news: you don’t need a massive rollout. You need a few practical moves you can start this week. Here are practical recommendations to help your team build momentum.

1. Use AI visibly, so your team knows it’s expected

If your team never sees you using AI, they’ll assume it’s optional or risky. Silence sends a signal.

I hear leaders say they want adoption, but when I ask how they use AI daily, there’s a pause. Some only tried a demo. Others use it quietly. The message their teams receive? If the boss isn’t using it out loud, maybe I shouldn’t either.

Insights from Atlassian’s AI Collaboration Index shows that when leaders model usage, teams experiment more. Psychological safety research echoes this: people take cues from what leaders do. So don’t just encourage adoption – show it.

Narrate your use in the flow of work. For example: “I used Rovo for the first draft.” “This agenda came from last week’s action items.” “AI surfaced an insight I might have missed.” Small, steady signals can help make AI feel like part of the cultural operating system.

Then tie it to the outcomes your team already cares about: faster briefs, fewer review cycles, cleaner handoffs. When results improve and the process is visible, adoption stops feeling top-down and starts feeling like momentum.

Consider trying this:

In your next one-on-one or team meeting, ask each person to share one way they’ve used AI in the past week, even if it didn’t work. Normalize the conversation. The goal isn’t perfection; it’s making AI usage something your team talks about openly, not something they wonder if they’re allowed to do.

2. Run a Fix-It-Friday to solve system problems together

Here’s a pattern I see often.

Five people on the same team are quietly fighting the same broken process. No one knows if the issue is the tool, the prompt, the agent, or the handoff.

Try running the Fix-It-Friday play. Get everyone together, pick one real problem, and spend 60–90 minutes diagnosing it. Ask:

  • Where is this actually breaking down?
  • Is it a system-level issue?
  • A prompting issue?
  • An agent configuration issue?
  • A human handoff issue?

When you do this together, you see what no one person can see alone. You also leave with one concrete change to your process – not just ideas that never land.

Speeding up individuals often exposes outdated coordination as the real constraint. That’s the AI efficiency paradox and Amdahl’s Law in action – and fixing that overhead is a team sport.

Consider trying this:

Add a recurring 60- to 90-minute block to your team calendar, monthly or biweekly. Bring one shared challenge to each session. Diagnose it as a group. Walk out with one change, not a brainstorm doc.

If you want a lightweight framework for running these sessions, the Atlassian Team Playbook has plays for retrospectives and health monitors that work well as a starting structure.

3. Clean up one data set, so AI has something real to work with

This is the unglamorous step that makes everything else work better.

AI is only as strong as the context we feed it. If data is spotty, tagging inconsistent, or taxonomy unclear, AI will summarize confidently – and get it wrong.

You see it fast. Ask AI for performance insights and you get numbers you can’t trust. Ask for a key account summary and half the story is missing because CRM notes are incomplete.

Atlassian’s 2025 AI Collaboration Index shows companies using AI to improve cross-functional collaboration are 1.8x more likely to report significant efficiency gains. That only works with clean, shared sources of truth.

Consider trying this:

Choose one data set that directly impacts your AI workflows. Campaign tags, opportunity notes, lifecycle stages, whatever causes the most friction. Then define three things: what “good” looks like (the allowed values and formats), who owns it, and which high-impact records to fix first.

You don’t need to boil the ocean. Fix 10 to 20 records. Create the standard. Train one team. Then expand. Good data hygiene compounds over time, and once AI has a clean foundation to work from, its outputs improve dramatically.

4. Turn your meetings into method-sharing sessions

Here’s the shift that ties it all together.

AI made individuals faster, and the bottleneck moved – to approvals, queues, handoffs, and mismatched methods. The State of Teams report found that 65% of knowledge workers say their processes and workflows don’t support collaboration well. And yet most team meetings are still stuck in the old pattern: checking status and moving tasks around.

As a manager, the highest-value thing you can do in your next retro, weekly, or one-on-one is shift 15 minutes from status updates to method sharing. Ask each person to bring one thing: a prompt that worked, a checklist they created, or a small process change that saved time. Capture them in a shared doc, so the whole team can reuse them.

Organizations focused on AI-enabled coordination are nearly 2x more likely to achieve org-wide efficiency gains than those focused only on individual productivity. If you want to see that at the team level, your meetings need to become the vehicle.

Consider trying this:

In your next team meeting, carve out 15 minutes. Ask everyone to share one reusable method: a prompt, a workflow shortcut, a handoff rule that reduced back-and-forth. Put them all on one shared page. Make it a living document. Come back to it every two weeks.

For a structured approach to running retros that surface these kinds of insights, try a sprint retrospective from the Atlassian Team Playbook.

Agility: The new workplace currency

Adam Grant made a point that stuck with me: the old currency of work was ability, your current skills, and expertise. The new currency is agility and your capacity to keep learning.

That resonated because it describes exactly what I’m seeing across teams right now. The leaders who are thriving aren’t the ones with the fanciest AI stack. They’re the ones who’ve made it safe to experiment, easy to share what’s working, and normal to say, “This isn’t working anymore, let’s adjust.”

You don’t need a massive transformation. Instead, develop:

  • Buffer time to try things
  • A ritual to fix what’s broken, together
  • Cleaner data so AI can actually help
  • Day-to-day transparency about how you’re using AI
  • Meetings that spread good methods, not just status updates

Pick one of these to try this month. Add buffer time to one project. Run a Fix-It-Friday. Then, make sure to label your next doc with how you used AI.

The point is to keep moving.

These learnings are just a snippet of insights shared from our recent Teamwork in an AI era event. Watch our sessions with Adam Grant, Forrester, and other researchers on-demand to learn more.