“Tag on LinkedIn when posting: Andrew Nusca, Caitlin Allen, Ben Gammell, Daniel Murphy, Stacy Simpson, and the company pages for Simbe Robotics, Brex, Liquid Death, athenahealth, Typeface.”
Last week in Aspen, five of us sat around a table at Fortune Brainstorm Tech to talk about the hard part of AI. The premise was simple: the technology mostly works now. The hard part is what happens after the demo, when you try to make it stick across a whole company. Roughly four out of five AI efforts stall before they reach real business impact, and almost none of the reasons are technical.
My own read, going in and coming out: individual AI productivity is largely a solved problem. One person can make themselves 10x more effective in a weekend. The hard part is making that gain compound across a team, then a department, then a company. Most of the stalled efforts are stuck in exactly that gap, and it is a people and process problem well before it is a technical one.
What stayed with me were the people, so let me hand the mic to them.
Caitlin Allen, who scales Simbe Robotics' shelf-scanning robots across 55+ retail banners, had the sharpest line on integration: "All processes, all teams need to be rooted in their own ways of operating so that this tool gets metabolized and absorbed." Every store is different. You earn scale by learning the business first.
Ben Gammell, President and CFO of Brex, named the tension underneath the hype: "We're actively on the cusp of being able to 10x the amount of quality code... what does that do to your customer?" Speed only helps if the rest of the system can absorb it.
Dan Murphy of Liquid Death gave the line everyone wrote down: "AI is democratizing mediocrity." When everyone can produce more, the moat becomes taste and judgment, not volume.
Stacy Simpson, CMO of athenahealth, brought the view from a regulated world: "In healthcare IT, AI isn't about what it adds. It's about what it removes." And the restraint behind it: "Just because you can doesn't mean you should."
And Andrew Nusca, who moderated, was a champ driving all of it. He kept pulling us off generalities and toward specifics, knew exactly when to follow a thread and when to move on, and made 70 minutes feel like 20. That is hard to do well. He did it well.
What struck me was how fast four very different businesses converged on the same point, across robots, fintech, beverages, and healthcare. All four were past the question of whether the model worked. The thing slowing each of them down was people, process, and integration. Ben and Stacy, both from regulated worlds, found that the governance they already had in place turned out to be an advantage for adoption.
My note on top of that: the teams that break through pick one painful workflow and wire it end to end, across infrastructure, coordination, and experience, then expand from there. Trying to transform everything at once or skipping any one of these three is how you join the 81%.
The technology is the easy part. Grateful to have shared the room.