On a recent call, the head of marketing at a mid-sized B2B SaaS company walked me through everything her team did to solve her business problem.
They had built a shared prompt library on Notion. They had released a branded voice guide. They would conduct AI introductory training twice. They held monthly office hours where the team’s most prolific AI users answered questions. The CMO personally wrote a memo modeling thoughtful use of AI and gently reminding everyone that the goal was substance over volume.
Yet the work continued to happen. Half-finished memoirs that read like the first drafts of something better. Slide plays that looked good on the surface and fell apart on the third ball. Copy of the newsletter that responded to the brief and missed the audience.
Workslop is the easy symptom to name. Diagnosing where this comes from is more difficult and lies at another level of the organization.
Where Most Workplace Conversations End
BetterUp Labs and Stanford Research — the original HBR study from September 2025 and a followed in January 2026 – put the numbers clearly visible. Forty percent of employees received work in the last month. Each instance cost just under two hours to clean. In a company of 10,000 people, this calculation amounts to approximately $9 million per yeargone, to fix AI-generated work that was supposed to save time.
The number that seems most difficult to me comes from Asana research on the state of AI in the workplace. Only 19% of knowledge workers say they are clear about what types of work AI should do in their role. This figure explains the rest of the data.
The dominant solution in the conversation at the moment is the one that my interlocutor has been using for six months. Leaders should model targeted use of AI. Teams must install clear guardrails. Individuals should develop what BetterUp calls a pilot mindset.
Human-AI output must meet the same standards as human-only output. That of Greg Kihlström recent MarTech article extends the argument into marketing-specific territory, asking marketing leaders to step up and define handoff lines with IT, legal, and procurement.
This is all correct. None of this is false. And all of this places the burden in one place: the individual instigator, the individual leader, the individual mindset. This is the layer most teams have relied on over the past 18 months, and the real fix lies elsewhere.
Your customers are searching everywhere. Make sure your brand introduces himself.
The SEO toolkit you know, plus the AI visibility data you need.
@media (maximum width: 768 px) { .headline-responsive { font-size: 30px !important; line-height: 1.3 !important; } }
Where the system is broken
Workslop is what happens when individuals produce AI-generated work and there is no connective tissue between them.
In a working marketing team, learning must happen quickly. The content specialist discovers within the first week that this model needs a longer brief and a more specific personality to generate something usable. The designer understands during the second week that this image tool wants the brand colors to be in hexadecimal and not plain English. The email marketer discovers in week three that the AI subject lines seem generic unless you provide it with the last three subject lines that worked.
Each of these learnings is real. Each was deserved. In most marketing teams I see, none of this knowledge goes very far. The content specialist doesn’t know what the designer discovered. The email marketer doesn’t know what the content specialist has learned. There is no shared space where someone says, “This is what worked, this is how I did it, try it your way and tell me what improves.” »
You end up with a team of competent people, each leading their own small R&D project in parallel. Everyone improves in their section. The team’s combined results always generate wasted work, because no individual learning ever reaches the next person. When someone changes teams, this learning disappears with them.
This is the part of the job scrap problem that no one clearly names. This is a learning coordination problem, and it cannot be solved by another training session or a more precise branded voice guide. This problem can be solved by building an infrastructure that enables learning between people.
What the AI Coordination patch looks like
In my book “Hyperadaptive,” I call this connection layer the AI activation hub. A hub is a small group of people within the organization, virtual, physical, or both, whose job is to maintain AI capabilities back and forth within the rest of the team.
This is different from a help desk, prompt library, or AI ticket inbox. These are static repositories that you access when you need to search for something. A hub is made up of people whose job is to actively move learning within the team.
The practical point of view, in a marketing context. A functional hub does a handful of specific things.
- Stay up to date and atomize learning: Hubs translate what’s new and what works into bite-sized, role-specific pieces of content that land in the workflow, looking more like a two-minute loom in a Slack channel than a wiki page that no one opens.
- Holds office hours and involves people: Hubs facilitate a convenient live experience. A hub member pairs an AI-savvy marketer with a marketer with a business background, and the resulting work is better than either could build alone.
- Maintains a usable knowledge engine: When engineering company iMBrace built theirs, they cut their information search time in half. This is the number to pay attention to. The repository is alive, searchable in natural language and constantly updated by the Hub itself.
- Measuring where AI earns and doesn’t earn: Hubs track what’s working and report that model to management. This is the element that is completely missing from most AI Marketing Centers of Excellence job descriptions.
When was the last time your team built something that intentionally allowed AI to learn between members, rather than hoping it would happen at the coffee machine?
New labor market data suggests that marketing is starting to understand this. According to a recent piece by Carilu DietrichMarketing AI leadership positions are growing rapidly under names such as Head of Marketing AI, Head of Marketing AI Center of Excellence, and Senior Director of AI Projects.
Related GTM engineer posts on LinkedIn more than doubled in six months, from about 1,400 in mid-2025 to more than 3,000 in early 2026. Marketing invents the hub in real time and gives it a different name.
The teams I see doing things right define the role correctly. They define the hub manager’s job around moving learning, matching people, and intentionally creating a smarter team. This is a different job description than “upholding AI standards and quality police prompts,” where most of these roles currently fall.
Where is this going?
The marketing teams that solve problems over the next 12 months will be the ones that build the connection layer that transmits learning between people. So when one marketer discovers something, the rest of the team uses it by the end of the week. This is the real solution. Efficiency follows. This is still the case.
The position Better Prompts Won’t Solve Your Work Problem appeared first on MarTech.
