Is Your AI ROI Getting Stuck In A Logjam?

According to the latest Atlassian State of Teams report, a mere 6 percent of executives are sure they can identify examples of organization-wide AI ROI. With 85 percent of knowledge workers using AI at work, many at their leadership’s behest, that’s a lot of activity churning out minimal results.

What’s happening? In my experience, many organizations jumped on the AI bandwagon before establishing a shared purpose. In their desire not to be left behind, they forgot to articulate the “why” for leveraging the shiny new technology. Unfortunately, when new imperatives are issued absent of a strategy – a shared goal, clear communication, and team supports – implementation is fragmented and reinforces the functional siloes that already exist.

Atlassian’s report backs this up: of those knowledge workers already using AI, only 29 percent of them have embedded it in their cross-functional workflows. As a result, individual execution is moving at warp speed while review / approval cycles struggle to keep up. This results in what Atlassian dubs the fragmentation tax. I call it the AI logjam.

So how do we make AI work for rather than against teamwork?

First, it helps to start with a plan. We think of it as “going slow to go fast.” Get clear on what problems you want AI to help solve, what tools will best serve this need, what guardrails need to be put in place (i.e., this tool can access proprietary info but this tool cannot), and what supports the team will need to adopt and integrate these tools effectively.

At my firm, we spent time last summer in a series of collaborative work sessions to explore what aspects of our business could be optimized using AI right away and the norms we wanted to put in place. This included time to experiment with tools and to share successes, frustrations, and questions. Approaching AI adoption as a team initiative, rather than an individual imperative, ensured that we were progressing in sync.

Second, I recommend looking at ways that AI can specifically support collaboration. Once a day, I leverage a Copilot agent to scan my email to determine if my lack of response is blocking progress for someone else. This way, I can stay largely focused on my priority work (and out of my email) without being a blocker. We also leverage Copilot within Teams to make sure our virtual meetings stay on track against the agenda and to create a recap of the conversation. Recaps are essential for making sure meeting action items and decisions aren’t lost in the ether. Now, they’re automatic.

Finally, a more general collaborative clean-up may be in order. Adding AI on top of a system that’s already broken is going to exacerbate challenges, not fix them. Moving quickly while still in alignment, regardless of AI, requires strong asynchronous collaboration processes, including document storage and versioning; norms for co-creation, reviews, and approvals; and knowing which communication channel (email, IM, meetings, etc.) to use when.

What the data shows is that leaving AI adoption up to individual initiative will not produce the desired returns. It creates more logjams than leverage. If you want to see meaningful gains, invest as much in orchestration as in the tools themselves.