To find AI use cases that work, start with the work employees hate

The best way to use AI right now is to let it handle the soul-crushing grunt work your team already hates, giving them more time for the work they actually enjoy.  

Companies are stuck in an awkward middle with AI. Everyone knows they should be using it, but most leaders are quietly trying to satisfy three conditions at once: 

1. Find work AI can actually do reliably 

2. Prove that it creates real business value 

3. Do all of this without alienating employees or stoking fears of layoffs 

On the surface, that feels like an almost impossible combination. But it’s only hard if you’re looking in the wrong place. The easiest way to satisfy all three conditions is surprisingly simple: start with the work employees already hate doing. 

AI excels at the tasks humans find most soul-crushing: summarizing endless documents, translating between languages, filling out forms, extracting data from PDFs and doing all of it over and over across thousands of pages. The great news is that your employees already hate doing this work. And they spend a meaningful chunk of their day on it: McKinsey has estimated knowledge workers spend about 20% of their time just searching for and gathering information, before they do any real work with it. 

This creates a natural division of labor. Leave the creative tasks to your human employees—they love doing that work and they do it far better than any AI. Let AI handle the boring, repetitive stuff that makes people watch the clock and dread Monday mornings. 

In fact, asking “what do people hate doing?” turns out to be one of the best criteria for identifying tasks that can be efficiently offloaded to AI. If a task is tedious, repetitive and nobody wants to do it, there’s a good chance AI can handle it — and everyone will be happier for it. 

The easiest way to understand where AI actually works is to start with something employees already hate doing. 

Let AI do the work people already hate doing 

Not the big judgment calls. Not approving mortgages. Not deciding investments. Just the boring, repetitive, soul-draining tasks that people rush through and do badly anyway. 

Think about contract tagging. Someone has to upload a contract and fill in 10 or 15 mandatory fields — company name, dates, jurisdiction, clauses. Nobody enjoys this. It’s Friday afternoon work. People rush it. They make mistakes. They click “accept” just to be done. 

Give that task to AI. Let it pre-fill the fields. A human still reviews it, but instead of two miserable hours, it takes five minutes. Low risk. Higher morale. Immediate productivity gains. Everyone wins. 

Or take mortgage applications. Humans waste time opening files just to discover something obvious is missing — a W-2, a pay slip, a required document. AI can spot that instantly and send it back without judgment. No risk. No decision-making. Just removing wasted effort. 

The same pattern shows up everywhere: sorting RFPs so they land with the right expert, summarizing long documents before humans read them, flagging incomplete files before anyone touches them. These aren’t new processes. They’re the same workflows — just faster, cleaner and less painful. 

One of the most telling moments came during a contract-tagging pilot. The assumption was that human-tagged data was “ground truth” and that AI was underperforming. Then someone actually checked the human work. 

It was terrible. 

Not because people were careless — but because they hated the task. And we have evidence that “manual, hated work” produces real error. A large systematic review in clinical research found manual medical record abstraction (humans extracting/transcribing fields from documents) had a pooled error rate of 6.57%— far higher than more automated or double-entry approaches. 

AI wasn’t worse than humans. It was better — because it doesn’t resent the work. 

That’s the real unlock. 

AI doesn’t replace judgment. It replaces drudgery 

And when companies start there — removing the tasks employees already dislike — you don’t get resistance. You get relief. No one feels threatened. No one feels replaced. They just get time back. 

That middle ground — between “no AI” and “AI decides everything” — is already big enough to deliver 10, 20, even 30 percent productivity gains. Not by changing who’s accountable. Not by rewriting processes. But by clearing away the work nobody wanted in the first place. 

That’s where AI actually wins today. 

This article is published as part of the Foundry Expert Contributor Network.
Want to join?