Better prompts won’t fix your workslop problem

On a recent call, the head of marketing at a mid-sized B2B SaaS company walked me through everything her team did to fix their workslop problem.

They’d built a shared prompt library on Notion. They’d published a brand voice guide. They’d run AI literacy training twice. They held monthly office hours where the team’s most prolific AI users took questions. The CMO personally wrote a memo modeling thoughtful AI use and gently reminded everyone that the goal was substance over volume.

Still, the workslop kept coming. Half-finished briefs that read like first drafts of something better. Slide decks that felt fine on the surface and fell apart by the third bullet. Newsletter copy that hit the brief and missed the audience.

Workslop is the easy symptom to name. Diagnosing where it comes from is harder and sits in a different layer of the organization.

Where most of the workslop conversation stops

BetterUp Labs and Stanford’s research — the original September 2025 HBR study and a follow-up in January 2026 — put the numbers in plain view. Forty percent of employees received workslop in the last month. Each instance cost just under two hours to clean up. At a 10,000-person company, that math comes out to roughly $9 million dollars a year, gone, to fixing AI-generated work that was supposed to save time.

The number that lands hardest for me comes from Asana’s State of AI at Work research. Only 19% of knowledge workers say they have clarity on what types of work AI should do in their role. That figure explains the rest of the data.

The dominant fix in the conversation right now is the one my call-mate has been running for six months. Leaders should model purposeful AI use. Teams should set clear guardrails. Individuals should develop what BetterUp calls a pilot mindset.

Human-AI output should meet the same standard as human-only output. Greg Kihlstrom’s recent MarTech piece extends the argument into marketing-specific territory, telling marketing leaders to step up and define handoff lines with IT, legal, and procurement.

All of this is correct. None of it is wrong. And every bit of it puts the burden on the same place: the individual prompter, the individual leader, the individual mindset. That’s the layer most teams pulled at for the last 18 months, and the actual fix lives somewhere else.

Where the system is broken

Workslop is what happens when individual people produce AI-generated work and there’s no connective tissue between them.

In a working marketing team, learning has to move fast. The content specialist figures out in week one that this model needs a longer brief and a tighter persona to generate anything usable. The designer figures out in week two that this image tool wants the brand colors in hex, not in plain English. The email marketer figures out in week three that the AI’s subject lines sound generic unless you feed it the last three subject lines that performed.

Each of those learnings is real. Each one was earned. In most marketing teams I see, none of that knowledge travels very far. The content specialist doesn’t know what the designer figured out. The email marketer doesn’t know what the content specialist learned. There’s no shared place where someone says, “This is what worked, here’s how I got there, try it your way and tell me what improves.”

What you end up with is a team of skilled individuals, each running their own little R&D project in parallel. Each one gets better in their slice. The team’s combined output still produces workslop because no individual’s learning ever reaches the next person. When someone moves teams, that learning walks out the door with them.

This is the part of the workslop problem nobody is naming clearly. It’s a coordination-of-learning problem, and it doesn’t get solved by another training session or a sharper brand voice guide. It gets solved by building infrastructure that carries learning between people.

What the AI coordination fix looks like

In my book “Hyperadaptive,” I call this connective layer the AI activation hub. A hub is a small group of people inside the organization, virtual, physical, or both, whose job is to keep AI capability flowing through the rest of the team in both directions. 

It’s a different thing from a help desk, a prompt library, or an AI ticketing inbox. Those are static repositories you go to when you need to look something up. A hub is people whose job is to actively move learning around the team.

The practical view, in a marketing context. A working hub does a handful of specific things.

• Stays current and atomizes the learning: Hubs translate what’s new and what’s working into bite-sized, role-specific content that lands in the flow of work, more like a two-minute Loom in a Slack channel than a wiki page nobody opens.
• Holds office hours and pairs people: Hubs facilitate live, hands-on experience. A hub member pairs a marketer with AI fluency with a marketer with business context, and the work that comes out is better than either of them could build alone.
• Maintains a usable knowledge engine: When engineering firm iMBrace built theirs, they cut their information-search time in half. That’s the number worth paying attention to. The repository is alive, queryable in natural language, and refreshed continuously by the Hub itself.
• Measures where AI is and isn’t earning its keep: Hubs track what’s working and surface that pattern back to leadership. This is the piece most marketing AI center of excellence job descriptions are missing entirely.

When did your team last build something that flowed AI learning between members on purpose, rather than hoping it would happen at the coffee machine?

The fresh job market data suggests marketing is starting to figure this out. According to a recent piece by Carilu Dietrich, senior marketing AI roles are growing fast under names like head of marketing AI, marketing AI center of excellence lead, and senior director of AI projects.

Related GTM engineer postings on LinkedIn more than doubled in six months, from roughly 1,400 in mid-2025 to more than 3,000 in early 2026. Marketing is inventing the hub in real time and giving it a different name.

The teams I see getting it right scope the role correctly. They define the hub lead’s job around moving learning, pairing people, and making the team smarter on purpose. That’s a different job description from “enforce AI standards and police prompt quality,” which is where most of these roles are landing right now.

Where this is headed

The marketing teams that solve workslop in the next 12 months will be the ones that build the connective layer that carries learning between people, so when one marketer figures something out, the rest of the team is using it by the end of the week. That’s the real fix. The efficiency follows. It always does.

The post Better prompts won’t fix your workslop problem appeared first on MarTech.