The short version: Workslop is AI-generated work that looks complete but has no real thinking behind it, and it's spreading through small teams faster than anyone's admitting. A 2025 Harvard Business Review study found 40% of employees had received it in the past month, and I've watched it happen inside my own client work. The fix isn't banning AI, it's putting a human checkpoint on anything that leaves your business with your name on it.
What workslop is (and why the name matters)
Researchers at Stanford's Social Media Lab and BetterUp coined the term in 2025 after surveying more than 1,100 employees across industries. Workslop is content that arrives looking finished, a report, an email, a strategy doc, a set of slides, but has no real substance underneath it. It reads well. It's formatted. And when you try to use it, there's nothing there.
The HBR piece put a number on the cost: employees who receive workslop spend an average of nearly two hours cleaning it up, clarifying it, or redoing it entirely. Scale that across a team of ten people and you're losing days every month to work that was supposed to save time.
That's the bit nobody wants to say out loud: AI hasn't made most small teams faster. It's made them busier producing things that then need a second person to fix.
A real example, because vague warnings help nobody
Last year I had a contractor send over a competitor analysis for a client pitch. Twelve pages, clean headings, a table comparing five competitors on pricing, positioning and tone of voice. It looked like exactly what I'd asked for.
Except when I sat down to use it, three of the five competitors had generic descriptions that could have applied to any business in that sector. The pricing data was three years out of date. The "tone of voice" column used the same three adjectives for every single competitor: modern, customer-focused, trustworthy.
It had clearly been generated in one prompt and sent across without anyone reading it. I spent an hour and a half redoing the parts that mattered, which was longer than it would have taken me to do the research myself from scratch. That's the trap. The document looked like it had saved time. It had cost it.
Why this is spreading faster in small businesses than anywhere else
Big companies have layers. A junior analyst's report gets reviewed by a manager before a client sees it. Small businesses often don't have that layer, which means AI-generated work goes straight from the tool to the client, the customer, or the decision that gets made.
I see three specific patterns that make small teams more exposed:
- One person wears five hats. The same person writing the marketing email is also doing the invoicing and the client calls, so there's no time to double-check AI output before it goes out.
- Speed is rewarded over substance. If you praise a team member for turning around a report "so fast," you're training them to prioritise speed, and AI is the fastest route to something that looks done.
- Nobody wants to be the one who says "this is thin." Calling out a colleague's AI draft feels awkward in a five-person team in a way it doesn't in a five-thousand-person one, so weak work slides through unchallenged.
This connects to something behavioural economists have written about for decades, and it's worth borrowing from the work of Richard Thaler on cognitive ease: things that look fluent and well-formatted get judged as more trustworthy, even when the underlying content is weak. Polished formatting is doing the persuading, not the thinking. That's exactly the trick workslop plays on all of us.
The uncomfortable part: it's not really an AI problem
Here's the bit that gets skipped in most of the coverage of this. Workslop isn't a technology failure. It's a standards failure that AI has made easier to hide.
Before AI, a lazy employee who wanted to fake effort had to type something out, which took time and left obvious gaps. Now they can generate four paragraphs of confident-sounding nonsense in ninety seconds. The laziness hasn't changed. The cover for it has got much better.
Which means the fix isn't an AI policy. It's a quality bar that existed before AI and should still exist now. Brands that win long-term treat that bar as non-negotiable, not as a nice-to-have. Coco Chanel refused to release a garment that didn't meet her own standard, even when it cost her money and time, and that discipline is exactly what's missing when someone forwards an AI draft without reading it. You can read more on how Coco Chanel built her standards into the business itself rather than treating them as optional polish.
How to spot workslop before it leaves your business
I now run every AI-assisted piece of client-facing work through the same five checks. It takes about ten minutes and it has saved me from sending out at least a dozen weak documents this year alone.
- Ask "what would I lose if I deleted this paragraph?" If the answer is nothing, the paragraph is filler and it needs to go.
- Check for specific numbers, names and dates. Workslop hides behind vague phrases like "many businesses" and "in recent years." Real thinking has specifics attached.
- Look for repeated sentence structures. If every section starts the same way or uses the same three adjectives, a human hasn't reworked it.
- Ask the person who wrote it one direct follow-up question. "Why did you choose this competitor over the other two?" A person who did the thinking answers in seconds. A person who forwarded AI output goes quiet.
- Time-box the review. Give yourself 10 to 15 minutes per document, no more. If it doesn't hold up in that window, it goes back, it doesn't get "fixed on the fly" under deadline pressure.
What to do instead of banning AI
Banning AI tools in a small business is pointless and I wouldn't recommend it. The answer is putting the effort into the parts AI can't do, and being honest with your team about where that line sits.
Want AI doing the heavy lifting in your marketing?
I build the systems that handle the boring 80 percent, so you get your week back. Done properly, with the human kept in.
I now treat AI output as a first draft from an intern who's fast but has no judgement yet. That means:
- AI drafts get a name attached to who reviewed them, not just who generated them.
- Anything client-facing gets one specific fact added by a human that the AI couldn't have known, a detail from an actual call, a real number from that client's account, a name of someone on their team. That one addition is usually enough to tell you whether real thinking happened.
- We measure output by what happened after it went out, did the email get replies, did the proposal win, not by how quickly it was produced.
This is the same principle behind why interactive tools tend to outperform generic AI content in marketing: a calculator or quiz forces something specific and useful into the interaction, whereas a wall of AI-generated text doesn't. I wrote about this in a piece on using interactive calculators to engage and convert visitors, and the underlying lesson applies here too: specificity is what makes people trust content, not polish.
What good AI-assisted work looks like
The brands that use AI well in their marketing don't let it replace their voice, they use it to speed up the boring parts while keeping the distinctive bits human. Look at how Headspace keeps a consistent, calm, specific tone across every channel, or how GoDaddy built a recognisable voice around small business owners specifically rather than generic "entrepreneurship" language. Neither of those brands could have built that with AI-generated first drafts left unedited. You can see the detail in how Headspace built its brand voice and how GoDaddy positioned itself around real small business owners. The AI tools existed for both of them by the time these strategies matured. The voice didn't come from the tool. It came from someone deciding what the brand was allowed to sound like and enforcing it.
Disney is the extreme version of this discipline, decades of consistency across thousands of touchpoints, none of it accidental. That's covered in detail in the piece on how Disney built a brand that wins, and the lesson transfers directly: consistency and quality are a decision you enforce, not a byproduct of the tools you use.
If your team needs this fixed
If you're running a small business and you suspect workslop is already inside your reports, your client emails, or your marketing copy, the fastest fix is usually not more tools, it's someone auditing what's going out and setting a standard your team can follow without you checking every document yourself. That's the exact gap an AI implementation coach is meant to close, building the review habits and prompts that stop this at the source rather than catching it after the fact.
I'll be blunt about the numbers here because vague reassurance doesn't help anyone budgeting for this: proper implementation support for a small team typically runs a few hundred pounds for a single audit and workshop, up to a few thousand for ongoing monthly support depending on team size. It's cheaper than the hours lost cleaning up workslop every single week, which the HBR research put at nearly two hours per document for the person who receives it.
Frequently asked questions
What is AI workslop, exactly?
Workslop is content generated by AI that looks polished and complete on the surface, correct formatting, confident tone, no obvious errors, but has little or no real thinking, research or substance underneath it. The term was coined in 2025 research from Stanford and BetterUp, published via Harvard Business Review, which found 40% of employees had received workslop from a colleague in the previous month.
Is workslop only a problem in big companies?
No, and it's arguably worse in small businesses. Large companies often have review layers, a manager checking a junior report before a client sees it, that catch weak AI output before it goes out. Small teams frequently don't have that layer, so workslop can go straight from an AI tool to a client or customer with nobody catching it.
How do I stop my team producing workslop without banning AI tools?
Set a quality bar rather than a tool policy. Require every AI-assisted document to have one specific fact, number or detail added by a human that the AI couldn't have known. Time-box reviews to 10 to 15 minutes per document, and ask the person who produced it one direct follow-up question before it goes out. If they can't answer it quickly, the work wasn't thought through.
How much time does workslop cost a small business?
The Harvard Business Review research found employees spend close to two hours on average cleaning up, clarifying or redoing a single piece of workslop they've received. Across a team of even five or six people, that adds up to several lost workdays a month, easily more expensive than the time AI was supposed to save in the first place.
Related reading: Why AI Is Quietly Wrecking Your Email Open Rates (And How to Fix It).
Related reading: AI Meeting Notes Are Quietly Killing Accountability on Small Teams and How to Use AI for SEO the Right Way (Without Wrecking Your Rankings).
