Most of the conversations happening today about AI focus on either end of the workplace spectrum: How the organization at large will adopt and implement AI for enterprise transformation at scale, or how individual leaders or workers are using AI to change specific habits and productivity.
Yet real change—and real adoption—happens somewhere in the middle. My conversations with executives show that the team is the unit of AI transformation, and one champion on a team is worth 10 mandates from the top.
Take a recent Charter Forum session we hosted, during which four companies shared what they’re learning about AI adoption. Their most consistent finding was that there’s a meaningful multiplier effect that happens when AI gets embedded into shared team rituals, workflows, and norms. Notably, all four companies arrived at the idea independently, each coming from different starting points and different frameworks.
At Charter’s Leading with AI Summit, meanwhile, Affirm chief people officer Barb Cadigan shared how her team puts this belief into practice. To meaningfully embed AI in engineering, her team didn’t launch a training program. It cleared the calendar. All 905 engineers on the team set aside everything outside of hiring and urgent bug fixes for an entire week to experiment and learn together.
I talk with a lot of leaders across companies, including in tech, who feel like the adoption metric they’ve been chasing has stalled and isn’t having an impact on broader goals. Utilization of tools is up, and individual output is growing, but it’s happening sporadically—and team goals aren’t really accelerating.
Below are insights about how to put the team at the center of your AI adoption efforts, drawn from my conversations with executives and experts, as well as from Charter’s Forum discussions. (Forum events operate under Chatham House Rule, but excerpts below are shared with permission.)
The plateau most organizations hit
When it comes to AI adoption, focusing on growing the number of individual users will only get you so far. The number of monthly active users (MAU) of AI tools often climbs, and then levels off.
One primary reason: not everyone is a builder. Builders will beat their head against the wall to master a new tool, take a product management approach to rebuild workflows, and keep experimenting past the point of frustration. The rest of the workforce isn’t necessarily resistant to using AI tools, but they are rational. They’re waiting to see whether AI is actually going to change how the team works, or whether it’s just one more thing to manage.
This is why having a champion on the team—someone who’s deep into AI and willing to share—helps move the team farther, faster. Upwork’s Cassie Veres, for instance, says the company’s efforts at increasing individual adoption weren’t having the impact it wanted. Instead, she says “we’re spending time with leaders to get them to make formal commitments within their team [to] evaluate their structures, embed AI, [and] encourage their teams to try things.”
She’s not the only one. All four companies in our Forum discussion said leveraging AI to improve shared workflows, reduce toil, and build momentum went further when smaller teams worked on them together. “The real value of AI comes at the team level,” says Atlassian lead behavioral scientist Ben Ostrowski. “Teams are where work has always been done.”
The cultural foundations of team-level adoption
To effectively get teams working on AI progress together, leaders need to make sure three elements are in place:
Shared norms. Talking about acceptable uses of AI within your team reduces people’s fear of being seen as incompetent and sets quality standards that reduce the odds of AI workslop. At Atlassian, teams that set explicit working agreements about AI use saw measurable results: AI use by non-power users went up by 10 percentage points, and team understanding of how AI could drive progress against goals jumped 82%, Ostrowski says.
Shared goals. Individuals who feel like they’re competing—whether to demonstrate personal adoption, hit usage quotas, or protect their jobs—don’t share what they’re learning. They hoard it. Kit Krugman, Foursquare’s senior vice president of people and culture, has said that after the company shifted from individual performance evaluations to team-based ones, teams added better practices and behaviors than if senior leaders had given them top-down mandates. You cannot build team-level AI norms on top of competitive individual incentives.
Embedded champions. Champions matter, but only when they’re operating at the team level—investing time with specific colleagues, setting norms, and automating the toil that belongs to that team’s workflows. Atlassian, for example, tracks which teams have a champion embedded as a leading indicator of meaningful adoption.
Udemy, meanwhile, organically surfaced more than 60 champions through its UDay program, an initiative that dedicates one meeting-free day per month to AI learning. The key outcome? Champions stayed connected to their teams rather than becoming a separate center of excellence, and increased active users of AI fivefold.
How teams help with burnout and ‘AI brain fry’
Shifting the focus from the individual to team-level adoption also reduces the impact of AI on burnout and what Gabriella Rosen Kellerman and her colleagues at Boston Consulting Group call “AI brain fry,” or the acute mental fatigue that comes from marshaling cognitive oversight of AI.
Participants in the Charter Forum session said giving champions the time and space to redesign workflows helped them reduce mundane, boring tasks for the entire team. Meanwhile, workers whose managers treat adoption as a team effort experience 15% lower mental fatigue, according to BCG. Workers left to figure out tools alone face what Kellerman calls the “AI orphan tax:” a 5% fatigue premium for going it alone.
“Being there to help humanize the experience of work, to help make it feel like a collective effort—that’s a huge part of what managers should be doing right now,” Kellerman told me.
Investments that accelerate impact
The shift from individual to team-level AI adoption requires some deliberate moves. Here are five things you can do now to make real progress:
Protect time for team-based learning. Self-paced modules and personal exploration don’t create the social reinforcement that changes how a team operates. Udemy gave teams a specific problem to solve within their functional unit, then shared results with other teams on Slack. Sharing what they learned in a public forum drove adoption in ways individual learning paths hadn’t managed.
Audit your metrics. If your primary AI adoption measure is how many monthly active AI tool users you have, you’re measuring activity, not outcomes. Add team-level indicators—which teams have an embedded champion, which have working agreements in place, and where cycle times or output quality have measurably shifted. At Charter’s AI Summit, Microsoft’s Katy George captured the shift: “We used to pay attention to adoption, now we just pay attention to performance.”
Set working agreements before you set usage targets. Give teams structured time to answer three questions: When should we use AI? When shouldn’t we? What does good output look like for our work? Teams that answer these together set norms that stick, and set the stage for deeper questions, such as how AI can give the team new paths for growth.
Align on goals before deploying more tools. If your cross-functional project teams can’t articulate a shared goal, no AI tool will save them. Before kicking off any initiative, invest in goal alignment first. Ask every member of the team to write down, in one sentence, what the team is trying to achieve. The results will tell you what you need to work on before you introduce anything new.
Ask leaders to be role models. Ask every people manager to do one visible thing with AI in front of their team next week. Don’t just present about using AI; actually build or create something in front of people. The permission that unlocks adoption comes from watching someone you respect look uncertain and try something new anyway.
Read more from Charter about AI adoption:
- Why four tech companies say adoption is the wrong AI metric
- How mastering hybrid is an AI advantage
- How HR leaders are moving beyond AI’s ‘efficiency’ metric