Your Team Is Saving a Day a Week With AI. And Hiding It From You.

Three new studies landed this week, and together they tell one uncomfortable story: your people are getting massive productivity gains from AI at work, feeling guilty about it, and quietly deciding not to tell you.

Forbes rounded up the research on July 13. BCG found that 42% of employees now save a full workday every week using AI. A workday. Every week. And 66% of them have received zero guidance on what to do with that reclaimed time. Employment Hero found that 42% of AI users feel like they are “cheating” when they use it, and 63% spend more time double-checking AI outputs than they expected.

Then there is my favorite new word in workplace research: botsitting. Glean found that supervising AI output now costs employees 6.4 hours a week, and 82% admit to delivering AI work they never verified.

Put the three studies side by side and a picture emerges that should keep every executive up at night. Not because the technology is failing, but because the culture around it is. Adoption is winning. Trust is losing.

Shame Grows Where Leadership Is Silent

When nearly half your workforce feels like using the company’s own tools is cheating, that is not a technology problem. That is a culture problem. Nobody has stood up and said: this is how we work now, here is what good looks like, here is what we do with the time we save. So people fill the silence with guilt, and guilt makes people hide things.

And hidden productivity is dangerous productivity. The 82% shipping unverified AI work are not lazy. They are unled.

Compare this with how leadership usually rolls out software: a license, a login, a lunch-and-learn, done. AI at work is not that. It changes what a workday is, so it deserves an actual conversation about norms, credit, and quality. Silence is a policy too, and right now silence is the default policy in most companies.

The BCG research makes the stakes plain: employees are quietly banking a full day per week while two-thirds navigate that windfall with zero direction. Guilt plus silence produces exactly the behavior leaders complain about most: hidden workflows, unverified output, and a shadow productivity economy humming underneath the org chart.

The Freed Workday Is a Leadership Test

Forty-two percent of your people just found an extra day in their week. What happens to that day is the most consequential unmanaged decision in your company right now. Does it go to deeper client work, learning, mentoring, building? Or does it evaporate into more meetings and quiet afternoons, while your best people wonder why they should mention the gain at all?

The Botsitting Budget

Now add the supervision tax. Glean’s 6.4 hours a week of botsitting amounts to nearly a full workday spent checking the machine’s homework. So the honest math of AI at work reads like this: save a day, spend most of a day verifying, and feel vaguely guilty about both. No wonder most users report spending more time checking outputs than they ever expected.

Smart leaders flip that tax into an asset. Treat verification as a named skill, write it into job descriptions, and score it in reviews. The moment checking AI work becomes recognized work, the shame evaporates and the quality problem shrinks with it. For example, one standard travels everywhere: nothing ships to a client until a named human owns it. Simple, public, enforceable.

Budget for the learning curve honestly as well. The 63% who spend extra time checking outputs are not failing at AI; they are climbing the trust curve every profession climbs with a new instrument. Pilots did not trust autopilot in year one either. Give your people six months of sanctioned practice, and the verification hours fall as pattern recognition rises. Punish them for the learning curve, and they will simply stop telling you they are on it.

Three Moves to Lead AI at Work This Quarter

First, say the quiet part out loud: AI at work is expected, not cheating. Second, set a verification standard so botsitting becomes quality control instead of anxiety. Third, decide as a leadership team where the reclaimed time goes, and reward the people who surface their gains instead of burying them. Trust is built exactly here, in the gap between what your people are doing and what they feel safe telling you.

There is a hiring angle here too, and it is a big one. The candidates who thrive in this environment show their process openly: their prompts, their verification habits, their judgment about when not to use the machine at all. That transparency is precisely what I screen for now in executive searches, because a leader who hides their process builds teams that hide theirs.

And a word to the 42% who feel like they are cheating: you are not. You are early. Every technology that mattered felt like cheating to its first honest adopters. Calculators felt like cheating. Spreadsheets felt like cheating. Meanwhile, the people who leaned in built the next version of every profession. So lean in, verify your work, and say what you did out loud.

The companies that win the next five years will not be the ones with the best models. Everyone rents the same models. They will be the ones where using the tools is normal, talking about them is safe, and the freed workday lands somewhere deliberate. That is not a technology strategy. That is a trust strategy wearing a technology costume, and trust has always been the scarcest resource in every workplace I have ever walked into.

Source: Forbes, July 13, 2026.

What the Best Teams Do Differently

The teams getting this right share four habits, and none of them require a single new tool. First, they publish norms in writing: where AI use is expected, where it is forbidden, and where judgment calls go to a named human. Ambiguity is the tax everyone else keeps paying. Second, they build a shared prompt library, because the productivity gap between a good prompt and a bad one is enormous, and hoarded prompts are just hidden workflows with better branding.

Third, they run show-your-work rituals. Once a week, someone demonstrates exactly how they used the tools on a real deliverable, mistakes included. The demonstration does two jobs at once: it spreads technique, and it makes AI use visibly normal, which drains the shame from the room faster than any policy memo. Fourth, they route the reclaimed time deliberately. A team that saves forty hours a week decides, out loud, where those hours go: deeper client work, a backlog project, learning blocks. Time that is not assigned evaporates, and evaporated gains breed cynicism.

Notice the pattern underneath all four habits: transparency, made structurally easier than secrecy. That is the entire game.

A Manager’s Script for the AI Conversation

Because so many leaders tell me they do not know how to open this topic, here is the meeting script, free of charge. Start with amnesty: “I assume everyone here uses AI more than I know, and that is good news, not a violation.” Say it exactly that plainly, because your people are calibrated for punishment and will test your sincerity for weeks.

Then ask three questions and actually write the answers down. Where is AI saving you real time today? Where does it produce garbage that costs you time? What would you do with a freed day per week if nobody judged you for having one? The answers will redesign your team’s workflow better than any consultant deck.

Close with one commitment and one standard. The commitment: nobody loses standing here for working smarter. The standard: nothing ships to a client without a named human owner. Repeat both every month until they sound boring, because boring is what safety sounds like.

The Career Stakes Nobody Is Saying Out Loud

Under the culture story runs a career story, and it is moving fast. Within two years, I expect “show me how you work with AI” to be a standard interview request at every serious company, the way portfolio reviews are standard for designers. The professionals building visible, verifiable AI practice today are quietly compounding an advantage that will look unfair by 2028.

The inverse is also true, and harsher. The employee who hid their gains, banked the quiet afternoons, and never developed verification judgment will interview like someone who spent two years standing still, because professionally, they did. Hidden productivity does not build a reputation. It builds a gap in your story that gets harder to explain every quarter.

For leaders, the same clock is running at the organizational level. Teams learn faster than policies, and your competitors’ teams are learning in the open. A company whose culture rewards transparency is compounding technique across every employee, every week. A company whose culture punishes it is training its people to improve alone, in secret, on skills they will eventually sell to someone else. The exit interviews will be polite, and the pattern will be invisible until it is a trend line.

So treat this moment as the hiring and retention event it actually is. The trust you build around these tools in the next two quarters will decide who is still in the building, and how good they have become, when the market turns.

AI at Work FAQ

Is using AI at work cheating? No. Cheating is claiming judgment you did not exercise. Using a tool to draft faster, then applying your expertise to verify and improve the output, is simply what working looks like now, the same way spreadsheets replaced adding machines without anyone calling accountants frauds.

Should employees disclose when they use AI? Inside the team, yes, and leaders should make that disclosure cost nothing. Externally, follow your industry’s rules and your client agreements. The dangerous zone is the current default, where everyone uses the tools and nobody admits it, because hidden workflows cannot be quality-controlled.

How do we verify AI output without losing all the time we saved? Tier it. Low-stakes internal drafts get a skim. Client-facing work gets a named owner and a real review. Regulated or high-consequence output gets structured verification with a checklist. Botsitting becomes manageable the moment it stops being one undifferentiated anxiety and becomes three explicit service levels.

Cathy Trinh is the Founder and Editor-in-Chief of Recruiter Hustle, OC/LA’s no-filter media platform for talent, finance, and recruiting professionals.

Heart. Human. Hustle.
Cathy