There is a telling bit of theater happening inside a lot of companies right now: leaderboards celebrating whoever burned the most tokens. It has the faint ring of transformation until you ask what actually got reinvented, and the room goes quiet. Don Scheibenreif calls that a lazy metric, and he would know—he spent fifteen years at Gartner building the machine-customer research and leading its Autonomous Business work. His read on the moment is that adoption is bottom-up while denial is top-down. His neighbor uses AI harder than most enterprises do.
Over ninety percent of CEOs use these tools daily. But when Gartner asked those CEOs who the most AI-savvy person on the leadership team was, the CISO and CIO topped out at forty to forty-five percent—nobody cleared half—and the chief HR officer came in last. That is the function that owns reskilling and job redesign, ranked least equipped, exactly when agentic AI stops being an IT project and becomes a question about how work gets divided. Don’s prescription is posture rather than budget: use the thing, ask good questions, and let your people see you use it. Do not just tell them to.
When Gartner started focusing on digital business in 2014, clients asked whether they were just describing e-commerce; the real prediction was disintermediation, and it only half landed. It lands properly now, because rebuilding what a middleman does got cheap and, as Don puts it, people want experiences that are intuitive, seamless, frictionless. He spent hours failing to reclaim a phone number he had held for fifteen years and simply gave up. That is the opening. The counterweight is trust: it took us the better part of fifteen years to stop arguing with Google Maps. Roughly seventy percent of people say they want shopping agents while under ten percent use them. Target is putting its own agent mistakes on your tab, eBay is barring agents—walled gardens like these will stall machine commerce until platforms let humans and machines actually meet. The strangest part is how little of this is new to him. Don sat down with Robb and guest host Michael Salamon in 2023 and reminded that Siri and Alexa do not work for the people using them. Three years later the warning is repeated. Our retrospective reads the two conversations side by side. We’ve also got an article by Sirte Pihlaja, whom Don name-checks in the episode, and a members article about the uneasy collaboration between writing and AI-powered tools.
Selling to Machines
Josh Tyson and Robb Wilson with Don Scheibenreif on bottom-up adoption and top-down denial, the executive AI savviness gap, disintermediation round two, customer effort as the ultimate service metric, and why walled gardens stall machine commerce.
They Still Don’t Work for You
Don Scheibenreif’s 2023 appearance with Robb Wilson and guest host Michael Salamon, read against his 2026 return—what changed, what held, and the question he asked three years ago that still has no answer.
Broken Telephone: Why AI Fixed the Wrong End of the Conversation
Sirte Pihlaja—one of the researchers Don Scheibenreif name-checks in this episode—on the strange asymmetry of everyone building AI to answer customers and almost no one building AI that calls on their behalf.
Usurper, My Collaborator
Josh Tyson on collaborating with Claude to pitch a novel, the one-minute pause before hitting return, and what happens when the coworker is built to replace you.
Frankle’s grid metaphor is the design brief: not another benchmark, but wiring that connects intent to verification. The 2023 mixologist and the 2025 fusion engineer are the same person asking the same question—what does success look like before you scale?