Have You Heard About Botsitting?

Six hours and twenty-four minutes. That's how much time the average knowledge worker loses every week to what researchers are now calling bot sitting — supervising, correcting and cleaning up AI output. The same workforce is saving a full working day a week through AI. The difference between teams that hold onto those gains and teams that trade one workload for another is deceptively simple: a framework for what to do with the time AI creates.

I've been living this. In the last few months I've built and rebuilt this entire website with Claude — a change impact scorecard, a change action plan, a combined tool that replaced both, then an AI readiness scorecard to sit alongside them. Numerous blog posts. Three full colour overhauls. A redesign that got redesigned. Work I genuinely couldn't have done alone, faster than I could have imagined. I've also stayed up until midnight more nights than I'd like, watching things process, wondering whether I should have brought my knitting. That feeling has a name now.

What the research says

Three studies published together in Forbes in July 2026 put numbers to something change managers have been seeing on the ground for a while.

First, the time opportunity. 42% of AI-using employees save a full working day every week — a genuine productivity shift, not a marginal one.

Then the guidance gap. 66% of those same people received no direction on what to do with the time they freed up. Reclaimed time doesn't naturally flow into high-value work without that direction. It fragments, and some of it flows straight back into the tool as verification overhead.

Then the confidence picture — and this is where it gets interesting. 42% of people using AI feel guilty about it, like they're cheating. Meanwhile, 60% of managers believe their teams view AI positively. Two completely different pictures, inside the same organisations. And 63% of AI users say the tool actually creates more work through the quality-checking it demands. 82% of heavy users have submitted output they didn't fully verify.

Put those numbers together: AI is generating time. Teams are losing most of it back through a lack of structure — no guidance on what to do with freed hours, no verification habit, no signal that it's genuinely safe to use the tool openly. The net result is 6.4 hours of weekly botsitting running almost directly against the gains.

Where to start

The teams that get the most from AI aren't necessarily using better tools. They have four things in place before they scale: governance (clear rules on what goes in and who reviews what comes out), transparency (people know what's changing and why), managers who can actually answer the questions their teams are asking, and training that builds real fluency — not just licence activation.

With those four things in place, the dynamic shifts. People use AI openly rather than quietly. Verification becomes a habit rather than a source of anxiety. The freed-up time flows where it should — and it compounds.

This is also where the perception gap closes. When leaders know where their people actually stand — not where they assume — they can target the right support at the right moment. The adoption curve accelerates from there.

The teams that keep the gains from AI are the ones who built the conditions for it. The framework comes first.

The Change Made Simple AI Readiness Scorecard was built from a change management perspective — which means it looks at the four areas that actually determine whether AI lands well in an organisation, not just whether the technology works. Ten questions, about 4 minutes, no sign-up. You'll walk away with tailored actions designed to give your AI project the best chance of succeeding.

Written by Sheena Karim Connect on LinkedIn ↗