Seventy-five percent of workers say AI made them more productive. Thirteen percent say it meaningfully improved their organization’s performance. I’ve been chewing on that gap all week, and I think it closes.
Rebecca Hinds, PhD, laid it out in Inc. Magazine, off the Glean Work AI Index from the institute she founded. Six thousand digital workers, eleven hours a week automated, and the saved hours get eaten by what she calls coordination neglect, showing up as handoffs, rework, and the six and a half hours a week she calls botsitting.
Coordination is one reason of several. Most look temporary.
Start with me. I’m a solo AI-native corporate-commercial lawyer, so no handoff. When I cut the time on a first draft nobody downstream redoes it, and no committee decides if it counted. The ~5x boost AI and automation gave me held. Sample size of one, true. But multiply me by six thousand and the real story looks downright frightening.
And ready or not, that’s where we’re heading.
Not every worker reports the savings. If a day’s work takes two hours instead of eight, there’s an incentive to keep quiet. And worker bees know what happens if they tell the boss: headcount drops, they pick up the slack, and the newfangled work-life balance goes poof. Self-reports run both ways. Hinds cites a 2025 trial where experienced open-source developers believed AI made them 20 percent faster but were actually 19 percent slower.
And most people aren’t using AI properly. Plenty of law firms banned it in 2023, and an unfamiliar tool in another tab doesn’t survive a busy Tuesday.
I read the botsitting number as a phase to break through. That is the price of a half-configured system. The trap is paying it weekly and calling it overhead. That’s bailing, not patching. Every hour of it points at something fixable, a bad prompt, missing context, or a job it’s bad at. Patch what keeps recurring and it stops recurring, easier when you built it than when you submit a ticket. Mine went from eating the gains to catching the few errors it still makes as I work.
The gains hold. I promise you.
None of that happens in a trial period, which is where people quit. Collapsing that work takes bona fide management and considerable patience, and lawyers are accidental managers. Until now nobody outside the profession could evaluate legal work (that’s changing, I built Ariadni to do just that), so voilà, you get lawyers who ‘manage’ with no real training whatsoever. And patience has never been a lawyer’s strong suit.
I ran teams before law school and majored in Industrial and Labor Relations, so the managing part came easy. Happenstance, not a virtue.
Hinds’ point stands and I’m not displacing it. It’s just not what’s stopping the lawyers I talk to. Most never got the tooling running. The gain isn’t vanishing in the handoffs, it never showed up. And I bet that’s happening to a whole lot of other orgs.
Attorney Advertising. Prior results do not guarantee a similar outcome. Not legal advice.
