The Adoption Gap is a fortnightly newsletter for leaders who want AI tools that actually work. Each issue cuts to the research, data and trends that matter, not the hype, not the doom, just what closes the gap between AI tools deployed and AI tools adopted. Curated by Mark Phelps and Drew Horton. A Mark Phelps production.
Last issue I wrote about self-calibration, and why the people who get the most from AI tools are the ones who know their own limits. This week I want to zoom out from the individual team member to the whole initiative.
This issue: ADOPT · the model
THE ONE TO READ
If you only read one AI article before your next leadership meeting, make it this one.
The budget’s approved and the demo went well. But months later the tool’s icon is still sitting in the toolbar, unopened. That’s probably not just bad luck.
The state of AI in 2026: On the road to ROI by Dan Tinkoff, Lieven Van der Veken, Michael Chui and Tara Balakrishnan, McKinsey (Aug 25, 2026) | FREE
Reading the research on AI rollouts, it’s plain to me that many stalled initiatives were decided before they shipped, usually on one or two of five questions that could’ve been asked. Here they are, in the order I’d ask them:
A · Alignment. Is your AI initiative aimed at a real business need, not at satisfying a leadership directive or keeping up with competitors? If the honest answer is “someone senior asked,” better to know that now.
D · Design. Is it shaped around the people and their workflows, with the planning and support to keep it running, or just deployed and left? I think it’s the most important of the five, and it’s the one with the hardest number behind it. In McKinsey’s 2026 survey, nearly three-quarters of the high performers (the 6 percent seeing real earnings impact from their AI investments) report fundamentally redesigning their workflows. Among other firms, it’s about one in four.
O · Ownership. Does one leader own the outcome, the milestones and the measure of success, or was it handed to Product or IT to figure out? Shared ownership tends to mean nobody has it.
P · People. Is your team helping shape the tool and the work around it? Are they engaged enough to want to use it? Teams want a say in what’s being done with them, and they rarely adopt what was done to them.
T · Trust. Ultimately, will your team rely on it for real decisions, or find ways to work around it? It’s a common consequence of the People axis above, and quietly shared workarounds are usually the first sign.
A word on that Design number, because we should be clear about what it says. Most of the high-performing firms report redesigning their workflows, and that’s a pattern, not proof of cause. The answers are self-reported, and this survey can’t tell us whether redesign produced the returns or whether the companies already earning them had the appetite to redesign. I’d still suggest redesign, but I’d hold the evidence loosely. The Harvard paper below looks at the flip side: what happens when the tool arrives and the work around it stays put.
LEADERSHIP LOAD OUT: BRING THIS TO THE MEETING
Which one or two would you bet your own initiative is weakest on? I’d pick before your next steering meeting, then ask someone closer to the work the same question. If your answers differ, that gap is the first thing to look at.
ALSO ON OUR RADAR
IBM Study: CEOs are Reshaping C-suite Roles for the AI Era by IBM Institute for Business Value (May 4, 2026) | FREE
In a survey of 2,000 CEOs, 83 percent said success with AI tools depends more on people’s adoption than on technology. Yet they said only 25 percent of their workforce uses AI tools regularly as part of the job, even though 86 percent believe their employees have the skills to work with AI tools. That’s the People axis, in the CEOs’ own numbers.
The State of AI in the Enterprise 2026: The Untapped Edge by Deloitte AI Institute (Jan 21, 2026) | FREE
Of 3,235 business and IT leaders surveyed, 37 percent said they’re using their AI tools at a surface level, with little or no change to the processes underneath. Only 30 percent are redesigning key processes around them. A different survey, a similar picture.
Artificial Intelligence in the Firm: Bottlenecks in Software Production by Fiona Chen and James Stratton, Harvard University (Aug 4, 2026) | FREE (PDF)
Across 718 software firms, coding agents raised pull requests by 23 percent. The time to merge one rose from about 7 days to about 10.5, and resolved issues showed no statistically significant gain. The code got faster. The review step became the bottleneck. A working paper, not yet peer reviewed, and it covers coding agents specifically.
PARTING THOUGHT
A rollout seldom fails on launch day. It fails in the meetings before, when nobody asked who’d own it or whether the people doing the work wanted it or would even use it. By the time the tool ships, most of the answer is already in, but you just can’t see it yet.
Know someone rolling out AI tools at their company? Forward this to them.
Hit reply and tell me what you think. We read every response.
And if there’s something you think we should cover, send it over. We plan these issues as one connected argument, but the best suggestions come from readers, and we’ll give you props when yours lands.
Published for leaders putting AI tools to work in the US and UK.
Mark Phelps
A Partner for Product & AI Integration Design
Mark Phelps is a product and design leader with more than 20 years of Fortune 100 experience at GE, Fidelity, Bank of America, TIAA and Marsh McLennan. He works as a fractional product design and AI integration partner, most recently at Rockefeller Capital Management and on a regulated AI-powered healthcare software product. Stanford AI Certificate, UI/UX Design for AI Products, earned with distinction. MBA with Beta Gamma Sigma. Named inventor on a pending design patent for an AI-powered medical device interface. Marine Corps veteran.