AI and the end of busy work

The number stopped me cold. Seventy-six.

That’s how many days per year the average employee wastes on administrative tasks that produce zero value. Research across multiple industries found that 26% of every workday disappears into managing email, processing expenses, coordinating business travel, and other tasks that exist only because no one has eliminated them yet. More than two hours daily. Gone. (Two months of working days. Imagine telling someone they will spend a third of a quarter doing nothing that matters.)

AI can fix most of this right now. Not someday. The technology works, the ROI is documented, and the tools are available today. Even basic prompt engineering skills can eliminate hours of repetitive formatting and drafting work.

And yet. Only about 5% of companies are generating value from AI at scale, per Stanford HAI’s AI Index and related enterprise surveys. The rest are stuck in pilot mode, or they’ve deployed AI tools while leaving administrative overhead intact.

Why does busy work persist?

The scale of the problem isn’t subtle. This drives me a bit mad whenever I see it on slide three of an ops review. A Kronos survey of 2,800 employees found 41% lose more than an hour daily to work-specific tasks unrelated to their core job. Another study showed workers waste six working weeks yearly on duplicated admin work and unnecessary meetings. Administrative tasks prevent 40% of employees from completing their core work, and nearly the same percentage regularly feel unhappy with the quality and quantity of their output.

You’d think eliminating this waste would be a no-brainer.

It’s not. Legacy systems and workflows outlast their usefulness by years. No single redundancy seems large enough to matter on its own, so leaders focused on today’s problems never step back to clear the accumulation. The result is a proper death by a thousand administrative cuts.

But there’s a deeper issue. Busy work provides something real work often can’t: visible activity that looks like productivity. Peter Drucker wrote about this in The Effective Executive (1966) and he was spot on, even 60 years ago. Being busy is not the same as being effective. Removing that visible activity forces uncomfortable questions about what people should actually be doing instead. The thing is, most organizations quietly decide those questions aren’t worth asking.

What AI actually eliminates

Let me be specific about what changes when you let AI handle administrative work. In building Tallyfy, I watched how many “core” responsibilities at mid-size companies turned out to be 80% formatting, copying, and chasing approvals once you looked closely.

A Harvard Business School study tracked 758 consultants. Those using AI completed 12.2% more tasks and finished them 25.1% faster. Quality improved too, with 40% producing higher quality results. The impact hit hardest for workers below average performance, whose output increased 43%. Even top performers saw 17% gains. Those are not rounding errors.

What disappears? Data entry. Report generation. Document formatting. Meeting summaries. Calendar coordination. Email drafting. Expense tracking. All the tasks that eat time without building any competitive advantage. Does removing them solve everything? No. But it clears the ground so you can see what actually needs doing.

The evidence from real organizations holds up. Kaiser Permanente physicians saved nearly 16,000 hours on medical documentation using ambient AI scribes. DLA Piper saved 36 hours weekly on content generation and data analysis. And Somerset Council employees gained 10 hours monthly, with 87% reporting positive benefits. The St. Louis Federal Reserve study found workers using AI saved 5.4% of their work hours. Organizations implementing AI-driven automation see productivity increases between 25-40%, with some reporting labor cost reductions up to 90% for specific administrative processes.

Hold up a second on this one. AI doesn’t just eliminate tasks. It forces a harder question about workflow design, because of what Ethan Mollick at Wharton calls the “jagged technological frontier.” Some tasks AI handles flawlessly. Others, seemingly just as routine, fall outside its capabilities. The Harvard study found this edge precisely: for tasks selected to be outside AI capability, consultants using it were 19 percentage points less likely to produce correct solutions compared to those working without it.

Take customer onboarding. AI can read contracts, create project spaces, set up billing, generate welcome documentation, and schedule meetings. But figuring out which contract terms need negotiation, or spotting unusual customer requirements? That still requires human analysis. You can’t just swap AI in for humans across the board. Workflows have to be redesigned around what AI handles versus what needs human judgment. An Inc. analysis of AI adoption patterns found that workflow redesign has the biggest effect on an organization’s ability to see bottom-line impact from AI. Not the tools. The redesign.

If your team is stuck here, Blue Sheen can help unblock you.

The gap between tools and change

Job displacement fears are escalating fast - what most leaders miss is the deeper workplace AI anxiety sitting underneath the headline numbers. Concerns about job loss due to AI rose from 28% to 40% in just two years, according to Mercer’s Global Talent Trends survey of 12,000 respondents. 62% of employees feel leaders underestimate AI’s emotional and psychological impact. I think this anxiety is largely rational, not some failure of imagination.

The numbers from Goldman Sachs’ workforce analysis stopped me cold: 46% of administrative work and 44% of legal tasks could be automated. Nearly half. Let that sink in. Around 25% of current work tasks globally sit in occupations exposed to generative AI, per a UN/ILO analysis. And fewer than 20% of employees have heard from their direct manager about the impact of AI on their job, per Mercer research. So companies face a choice: eliminate the busy work and confront the job security question directly, or maintain administrative overhead to avoid difficult conversations.

Most choose the path of least resistance. AI tools get deployed, but existing workflows stay intact. The thing is, this is mostly bikeshedding around tooling choices instead of touching the actual job design. The vast majority of companies have not redesigned processes based on AI capabilities. As HBR’s analysis of AI-driven process redesign makes clear, most organizations bolt AI onto existing workflows rather than rethinking them. Automation happens around the edges, but actual jobs never get redesigned. Teams end up with AI tools while still spending time on the same administrative tasks, because nobody officially removed those tasks from job descriptions. Productivity gains show up as people doing more total work, not better work. Most challenges in AI rollout relate to people and processes, not technical issues. Breaking this pattern requires leadership willing to redesign jobs around what humans do better than AI. Can technology solve this on its own? No. This is a leadership problem dressed up as a technology problem.

What comes after the busy work is gone

I waffle between two views here, because the data points both ways and I keep flipping on which one wins. The post-busy-work organization looks fundamentally different. Not because AI does administrative tasks. Because eliminating those tasks forces clarity about what humans should do instead. Actually, ‘forces’ is too strong. It creates the opportunity for clarity. Whether anyone seizes it is another question.

Start with role redesign for AI. When 26% of someone’s day opens up, what fills it? That answer determines whether any of this creates real value or just shifts workload around. In conversations I’ve had with ops leaders on this exact question, the plain answer is that they have not thought about it yet, and the calendar is already full. Companies getting this right restructure roles around three categories: work only humans can do, work AI handles fully, and hybrid work requiring both. The first category expands. Another disappears. The third becomes the new frontier where you compete.

New metrics for productivity follow. Traditional measures focused on output volume: emails sent, reports completed, meetings attended. W. Edwards Deming warned about exactly this. Managing by visible figures alone misses what actually matters. When busy work disappears, volume becomes a meaningless signal. What matters is decision quality, relationship depth, strategic insight, creative problem-solving. All the things that don’t scale through automation.

Cultural rollout is harder. Organizations built around visible activity struggle when that activity disappears. You need different signals for who contributes value, different criteria for advancement, different expectations for how people spend time. This doesn’t happen on its own.

The hardest part, probably, is acknowledging that some roles existed primarily to manage administrative overhead that AI now eliminates. The World Economic Forum estimates 120 million workers are at medium-term risk of redundancy because they’re unlikely to receive needed reskilling. The displacement comes first, and it is painful. New roles take longer to emerge and require different skills. Only about 30% of organizations using AI have been proactive in training employees to work alongside it, per SHRM research. That gap matters more than most organizations acknowledge.

Leaders serious about this face it directly. They identify which administrative roles disappear. People who can shift to higher-value work get retrained; those who can’t require difficult decisions. Compensation and advancement get redesigned around new definitions of productivity. Success stops being measured by how busy people appear.

The more I look at the gap between organizations that actually eliminate busy work and those that just buy more software, the clearer it gets. The dividing line is whether leadership will have the awkward conversation about which jobs are mostly busy work, or duck it and call the resulting half-step a strategic win.

Seventy-six days a year. That was the number from the opening. The technology to reclaim those days exists right now. What most organizations lack is the willingness to confront what people should actually do with the time.

Busy work is no longer a necessity. It’s a choice. The question isn’t whether you can use AI to eliminate it. The question is whether you’re willing to confront what comes after it’s gone.