the numbers don’t add up. enterprise ai spend is at all-time highs. seat counts on the major copilots are huge. and yet — talk to anyone actually inside one of those companies, and the honest answer is “we have it, kind of, but i don’t really use it for the work that matters.”
that’s the gap this episode is about. and it’s not an adoption problem. it’s a workflow problem.
10x employees, 1.1x teams
joel’s framing for what’s actually happening landed early and stuck:
“you have 10x employees and 1.1x teams. the individual gets faster with ai. the team around them doesn’t change at all. the gap is not at the desk. it’s between the desks.”
the seats got bought. the workflows didn’t get rebuilt. so one person on the team writes their emails 5x faster, drafts their decks faster, codes faster — and then hands the output into a process that wasn’t redesigned to absorb the new throughput. the bottleneck moves down the chain. the team output barely changes.
the company looks at the team and says “well, ai isn’t working.” what isn’t working is the assumption that buying seats is the same thing as adopting the technology.
the unsexy work nobody pitches
if it’s a workflow problem, the fix is workflow work. and that is exactly what no consultant and no vendor and no transformation deck wants to sell.
“the actual answer is the unsexy work of changing your own workflows. nobody pitches that. you can’t put it on a slide. you can’t sell a license to it.”
workflow redesign is consulting hours, not software seats. it’s a five-month engagement, not a quarterly renewal. it’s specific to your org. and it produces no logo on a customer slide for the vendor. so it doesn’t get sold. so it doesn’t get done. so the 1.1x stays 1.1x.
the consultant’s honest line
rodrigo pushed on a version of this that joel just had to admit:
rodrigo: “the vendors are a little bit too comfortable. they need us — the consultants. they need the workflow not to fix itself, because if it did, there’s nothing for the next slide.”
joel’s response to that, on air, is one of the lines we kept coming back to across the arc:
“if we’re lazy, we make more money. that is the honest version of the consulting business model around ai. if the work doesn’t get adopted, we keep getting called.”
that is the structural reason why “nobody is using ai” stays true even as the budgets keep climbing.
the better question
late in the episode rodrigo reframed the show’s title and improved it:
“it’s not ‘nobody is using ai.’ it’s ‘most organizations are not using ai properly.’ and that is actually an opportunity, not a complaint.”
that’s where the rest of the arc went. not “the technology doesn’t work” — the technology works fine. the question is what has to change in the operating structure of an organization for the technology to land.
next
the workflow gap shows up in your dashboards as “ai pilots that didn’t convert.” the version that shows up in your incidents is uglier. next episode is what happens when the ai actually does ship — and the operations layer underneath isn’t there.
episode three: ai fails in production, not in the pitch.