Co-authored by Gerard Pietrykiewicz and Achim Klor
Most of us have heard the pitch by now:
“We can replace this team with agentic AI.”
The spreadsheet says it saves money. Leadership pulls the trigger. Headcount goes down. The announcement goes out.
Then the work slows down. Quality drops. Customers feel it. The team that remains spends its days cleaning up mistakes instead of moving the business forward.
Efficient? Maybe. Effective? Not at all.
What you’ve actually done is delete the operating context your AI needs to work.
Most leadership teams miss the human reaction.
When employees see colleagues lose jobs under the banner of “AI efficiency,” they do the rational thing: they protect themselves.
They stop sharing how the work actually gets done. Edge cases stop getting flagged. Nobody says “this breaks every quarter because of X” anymore. Experimenting in public feels like a career risk.
That is exactly the knowledge AI needs to be useful. And it disappears quietly, before anyone notices.
“Context is everything.”
The cut happens in the boardroom. The damage shows up in support tickets, missed handoffs, bad data, and customer calls nobody prepared for. By the time leaders see it, the people who knew the work are already gone.
The tool may be capable. But the operating context has disappeared, and without it, AI gives you generic answers to specific problems with zero context.
That’s a leadership problem (not technology) and why most AI adoption stalls at exactly this point.
According to a February 2026 Careerminds survey of 600 HR professionals, 32.7% of organizations that conducted AI-led layoffs had already rehired between 25% and 50% of the roles they cut. HR Executive, citing Forrester’s 2026 Future of Work research, reported that 55% of employers regret laying off workers because of AI.
A U.S. Express Employment Professionals-Harris Poll survey, reported by HR Dive, found the average cost of turnover has risen to $45,236 per worker.
The rehiring cost is just the invoice. The reset is what kills momentum.
New employees don’t just learn the job. They learn your version of the job: the edge cases, the customer history, the “this looks clean in the system but it’s actually wrong” reality that never made it into documentation. Someone has to teach the AI system what good looks like in your specific workflows.
“Tribal knowledge doesn’t come with the technology.”
In B2B, context lives inside customer relationships.
What was promised. What went sideways last renewal. Who actually makes the buying decision. Which account needs careful handling. Which “small” issue could quietly put a six-figure contract at risk.
None of that lives in your CRM. It lives in the people who manage those relationships.
When they leave, it goes with them. Your AI doesn’t know the history. Your new hire doesn’t either. The customer on the other end notices before you do. A staffing decision in a spreadsheet becomes a revenue problem in the field.
Klarna found this out publicly. The company said its AI chatbot could handle the work of 700 customer service agents. Later, the CEO acknowledged the customer-service cost-cutting push had gone too far and said the company needed to invest more in human support.
McDonald’s learned it in a simpler workflow. It ended its AI drive-through pilot with IBM after mixed results, complaints, and examples of misunderstood orders. The demo looked clean. The drive-through did not.
Building an AI-capable team is harder than buying AI tools. You need people who understand the business and know how to use AI to move faster, analyze better, and handle the parts that don’t need human judgment.
When you lose someone with both company-specific knowledge and AI fluency, you take a double hit. You’ve lost the context holder and the person who could have pulled the rest of the team forward on AI.
And you’ve donated that investment to a competitor. You paid for the learning curve, the mistakes, the AI up-skilling. Someone else collects the benefit.
Demand for skills that explicitly reference AI grew 109% year-over-year, according to Upwork’s 2026 In-Demand Skills report. That talent is scarce. Once it walks, you’re behind on headcount and capability at the same time.
Turnover is the polite word for it. You built their capability and handed it to a competitor.
Counterintuitive as it sounds, if you’re serious about agentic AI, you may need to hold onto more people in the near term. The case for keeping people isn’t soft. It’s that your AI is only as good as the context you can feed it, and right now that context lives in people. The ones who know where the bodies are buried in your workflows are the same ones who can make AI useful.
The same leadership gap shows up in how companies handle AI adoption barriers: unclear expectations, no safe space to experiment, and people who go quiet rather than risk being wrong.
Before you automate or cut anything, ask three questions about every role you’re considering:
Pick one workflow where you think an agent could genuinely help. Have the people who do that work map the real process, including the exceptions. In plain language. In examples. In failure modes.
Start there. And when you do go back to hire, make sure you're hiring for the right reasons.
Agentic AI will change org charts. The companies that get ahead won’t be the ones that cut fastest. They’ll be the ones that kept the context, trained the people, and built the agents around them.
The ones that didn’t will spend the next two years rehiring and wondering why the tools aren’t working.
Announcing efficiency is the easy part. The hard part is showing up six months later when the work still needs to get done.
If you like this co-authored content, here are some more ways we can help:
Cheers!
Achim is a fractional CMO who helps B2B GTM teams with brand-building and AI adoption. Gerard is a seasoned project manager and executive coach helping teams deliver software that actually works.
This article is AC-A and published on LinkedIn. Join the conversation!
