Half-Finished AI Work Tells You Something About the Pressure on Your Team
David Buirs is a leadership and executive coach in Amsterdam. This article is for managers who want to prevent AI slop: work made with AI that looks polished and is only half done. You learn where it comes from, why it damages client trust, and which agreements and conversations help.
Friday, half past four. You open the client analysis Mark just sent. Twelve pages, clean headings, tidy paragraphs. You read it twice and still can’t tell what he recommends. It’s the third time this month. How do you prevent AI slop in your team without banning AI?
The short answer: you prevent AI slop with three things. A team that understands why it matters. A clear agreement on when work is finished. And a team that feels safe saying it doesn’t know something yet. All three are largely in your hands.
Back to that Friday. You feel irritation, and something underneath it. Did he even read this himself?
Then comes the thought you keep to yourself. If I rewrite this now, I’ll be done by eight.
What is AI slop, and why am I getting it from my team?
AI slop is work that looks polished while the substance is missing. Researchers call it workslop. Whoever receives it ends up doing the thinking.
The Stanford Social Media Lab and BetterUp Labs coined the term in 2025. In their survey of 1,150 US desk workers, 40 percent said they had received this kind of work in the past month. Each case took almost two hours on average to fix.
It also hurts the relationship. About half of those surveyed saw the sender as less capable and less reliable afterwards. These are US averages, and the frustration will sound familiar.
Why does my team produce AI slop?
According to the same researchers, workslop mostly comes from unclear AI mandates and overloaded teams. Kate Niederhoffer, Alexi Robichaux and Jeffrey Hancock wrote this in Harvard Business Review in January 2026. Many employees are exhausted and feel unsafe admitting uncertainty or asking for help.
That is the reframe of this article. AI slop often tells you something about the pressure on your team. It’s a signal, and signals can be read.
Under that pressure, four causes tend to show up.
The first is pace. When someone juggles five deadlines, AI becomes a way to hand something in on time. Finished starts to mean submitted.
The second is an unclear brief. Perhaps the company told everyone to use AI. Nobody explained what good work looks like once you do.
The third is uncertainty. Someone isn’t quite sure what the client needs. A long ChatGPT answer then feels safer than a short question to you.
The fourth is craft. AI text looks finished. People with less experience spot the gaps more slowly. Then the work is mainly a learning moment.
Why is sending AI slop to clients so damaging?
Because it costs trust, and trust is what your client relationships run on. A client pays for your team’s knowledge and judgement. When they notice they are getting unfiltered AI text, they start to wonder what they are paying for.
Colleagues already rated senders of workslop as less reliable. With a client, that weighs even more. They judge your whole team on that one document.
Your team is often less aware of this than you are. You are in the client conversations and you hear the criticism. Your people mostly see their deadline. That is why it helps to explain it out loud.
An example helps your team feel what it is like for a client. You can share it like this: “Imagine I read you a text in your performance review that was clearly written by AI. You’d notice within a few sentences. It would feel impersonal, as if I couldn’t make time for you. That is how a client feels when they get AI slop from us.”
Online, someone who passes on AI output without thinking is sometimes jokingly called a “meat proxy”. A human relay.
You can tell your team: “We hired you for your knowledge and experience. That is what makes our work good. Use AI as support whenever it helps. You always remain responsible for what you send. If all you do is copy and paste AI text, you make yourself replaceable.”
That last line sounds harsh. It is also a compliment. You are telling them their own thinking is essential.
How can I prevent AI slop in my team?
Build awareness, and make good agreements alongside it. Agree together on when work is finished, and make doubt welcome. Those are exactly the things the research says are often missing.
In your team meeting: “Using AI is fine, and I encourage it. Before you send anything, I want to see three things. Your conclusion at the top, checked numbers, and a text that is as short as it can be.”
Then make doubt explicitly welcome: “If you’re unsure what the client needs, send me a question first. That’s quicker for me than twelve pages.”
Look at the workload too. Ask in your one-on-ones how many deadlines are running at once. If it’s too much, drop something or move it later in the plan. A little air in the schedule gives people time to think for themselves.
And lead by example. Tell your team how you use AI, and where you choose to think it through yourself.
For managers who want to work on this in a broader business context, business coaching for managers is a logical step.
How do I address AI slop without judging?
Start with the work and with what you miss. Then ask how it was made. That keeps the conversation about quality, and your team member has nothing to defend.
About the document itself: “I’ve read your analysis twice. I see a lot of text, and I miss your advice. What would you recommend to the client, in two sentences?”
If you suspect pressure: “This looks like it was put together quickly. How full is your plate right now? I’d rather have a shorter piece that’s right.”
If you want to know how AI was involved: “I’m fine with you using ChatGPT. I’m curious how you used it. Which part did you check yourself?”
This keeps responsibility with your team member, and makes it safe to be honest about pressure.
How do I stop fixing the work myself?
Hand the work back, with clear feedback. Every time you rewrite it over the weekend, your team learns that half-finished is good enough.
Rewriting feels faster and safer. In Positive Intelligence, a voice like that is called a Saboteur. Here it is often the Controller. It wants things done right, and done now.
Just before you start rewriting, ask yourself one question. Am I fixing this document, or the pattern?
Do you have to rescue the piece anyway, because the client expects it on Monday? Then still talk it through: “This time I wrote the conclusion myself. Next time I want to see yours.”
What do I gain when I prevent AI slop?
You get work back with the thinking in it. Your team learns to use AI as support for their own judgement. And you prevent drama: damaged client relationships and lower quality work.
For your team, it means faster work that is right, and more ownership. Your clients keep trusting you. Preventing AI slop comes down to awareness, clear agreements and a team that feels safe. Then AI becomes an accelerator, and the thinking stays with your people.
If you want to explore how to approach this in your team, business coaching for leaders helps you see your own patterns clearly. For leaders at director level who set AI agreements for the whole organisation, executive coaching is a natural next step.
Interested or curious? Let’s chat. Plan your free introduction here.
