AI Saves Your Team a Day a Week. They Lose Most of It.

The headline number that AI vendors love: workers using AI save 11 hours per week. That is from the Glean Work AI Index, published June 2026, surveying 6,000 full-time digital workers across the US, UK, and Australia — including 1,500 Australians. Seventy-five per cent say AI makes them more productive. The number sounds like a win.

Then you look at where the time goes. Workers spend 6.4 hours per week on what the researchers call 'botsitting' — the hidden labour of making AI actually usable. Re-pasting documents and business context into prompts eats 2.3 hours a week alone, the single largest category. Then supervising outputs. Debugging mistakes. Rerunning prompts that produced confident-but-wrong answers. Cleaning up before anything is client-ready.

Net time saved: 4.6 hours. Not even a half-day. For every hour of useful AI output, your team spends another hour making it usable. Workers using four or more AI tools — and a third of those surveyed do — are 35 per cent more likely to report frequent botsitting. More tools, more babysitting.

BCG's fourth annual AI at Work report, surveying 11,749 workers across 14 markets, found that 42 per cent of regular AI users save a full day or more per week. But 66 per cent receive limited or no guidance on what to do with the time. More than half do not reinvest it into strategic work. The time AI frees up does not get captured — it leaks into more emails, longer meetings, low-value tasks that expand to fill the gap. We have written about how wrong metrics hide this problem. The BCG data explains why: it is not a measurement failure. It is a direction failure.

This shows up in the bottom line. Only 13 per cent of workers in the Glean study say their organisation is performing significantly better because of AI. BCG found that clear strategy lifts measurable business impact by 25 percentage points. Better tools alone lift it by five. The return on AI is not a technology question. It is a leadership question.

Take the loaded rate for a skilled tradesperson in Australia — $80 per hour including wages, super, vehicle, and insurance. Eleven hours of AI time savings is worth $880 per worker per week on paper. Subtract the 6.4 botsitting hours and you are at $368. Now apply the BCG finding that two-thirds of workers have no direction for the saved time. The actual captured value for the average worker lands around $125 per week. A 10-person team that nominally saves $8,800 a week from AI probably captures $1,250. The rest evaporates.

For professional services firms, the arithmetic is sharper at a $120 loaded rate. And the Glean data adds a second cost: 69 per cent of AI users admit to shipping work they have not fully verified, do not fully understand, or cannot confidently stand behind. In an industry where professional liability attaches to every piece of advice, unverified AI output is not just a quality problem — it is an insurance problem.

First, cut the botsitting. The single biggest time sink is re-pasting context into prompts — 2.3 hours per worker per week. Tools that connect to your business data (your CRM, job management system, practice management software) eliminate that repetition. If your team is copy-pasting the same customer details, job notes, or compliance checklists into AI prompts multiple times a day, you do not have an AI problem. You have an integration problem.

Second, direct the time. The BCG data is unambiguous: strategy lifts impact five times more than better tools. Tell your team specifically what to do with the hours AI frees up. For a trades business, that might mean using saved dispatch time for same-day quoting follow-ups. For an accounting firm, it might mean redirecting admin hours toward advisory conversations. The instruction does not need to be complicated. It needs to exist.

Third, measure net time, not gross. If your AI vendor says the tool saves 11 hours a week, ask what the botsitting cost is. Track total time spent on AI-related tasks — prompting, reviewing, correcting — alongside output metrics. The businesses in the 13 per cent are not using more AI. They are using it with less friction and more direction.

Key takeaways

Related notes

Want to close the gap between what AI saves and what your business captures?

A 30-minute call can identify where your team's AI time is leaking — and what to do about it. Book a call.