Kief Studio

Half of AI Layoffs Are Expected to Backfire. If Your Agency Gutted Its Tech Bench, You Can't Catch the Rebound.

Build

  • Published
  • Updated
  • 8 min read
Violet and gold neon circuit lines with glowing junctions, cover for a note on agencies that cut their tech bench

TL;DR: Forrester found 55% of employers regret AI layoffs and expects half of AI-attributed cuts to be reversed. The work comes back, often cheaper and outside the company. Agencies that kept technical people, in-house or through a build partner, can take that work now. Agencies that cut their own bench will struggle to hire it back.

AI layoffs were sold as a clean trade: fewer people, the same output, better margins. A year of data says the trade is messier. Forrester found that 55% of employers regret laying off staff because of AI. (Computerworld (opens in a new tab)) Not "might regret." Already regret.

I spend a lot of my time on AI adoption and governance, and the pattern I see is consistent. The cuts happen on a forecast. The work removed by AI layoffs does not disappear. Someone has to pick it up. For agencies, that someone could be you, if you still have the people who can do it.

Did half of AI layoffs really backfire?

The honest answer: a large share of AI layoffs are expected to be reversed, and many were made before the AI could do the job. Here is what the research says.

  • Half reversed. Forrester's "Predictions 2026: The Future Of Work" says half of AI-attributed layoffs are likely to be reversed, and that "much of this work will be given to lower-wage human workers, offshore or at lower salary." (The Register (opens in a new tab))
  • Rehiring by 2027. Gartner predicted in February 2026 that half of companies that reduced customer service headcount due to AI will rehire staff by 2027 to perform similar functions. (CX Dive (opens in a new tab))
  • Cuts on a forecast. A Harvard Business Review survey of 1,006 executives in December 2025 found that 60% of organizations had already reduced headcount "in anticipation of AI's future impact," as Sherwood News reported. (Harvard Business Review (opens in a new tab); Sherwood News (opens in a new tab))
  • Little measured effect so far. In surveys of nearly 6,000 executives in four countries, "more than 90 percent of executives report no effect of AI use on employment over the past three years." (National Bureau of Economic Research (opens in a new tab))

What the backfire headline gets right and wrong

It gets the direction right. Many AI layoffs were made on the promise of the technology, and Forrester says so directly: "Too often, company management lays off employees based on the future promise of AI." (Computerworld (opens in a new tab))

It gets the ending wrong if you read "backfire" as "everyone gets their job back." Forrester expects the work to return at lower pay or offshore. So the lesson for an agency is not that the old staffing model is safe. It is that the work outlived the cut, and it needs a home.

What happened at Klarna after its AI layoffs?

Klarna is the clearest public example of AI layoffs meeting reality. A year after claiming its AI chatbot could do the work of 700 representatives, the company turned back to people for more of its customer service, and said customers should always have the option to speak with a human. (CX Dive (opens in a new tab))

Its CEO, Sebastian Siemiatkowski, put it plainly to Bloomberg: "As cost unfortunately seems to have been a too predominant evaluation factor when organizing this, what you end up having is lower quality." (CX Dive (opens in a new tab)) The chatbot still handles a large share of conversations. What changed is the assumption that it could handle all of them.

I find that reassuring rather than embarrassing. It is what responsible AI adoption looks like when the data comes in: keep what works, bring people back where judgment matters, and say so in public.

Why AI agents struggle with the work that got cut

The research explains the gap. In CRMArena-Pro, a benchmark for business tasks, "leading LLM agents achieve only around 58% single-turn success," dropping "to approximately 35% in multi-turn settings." (arXiv (opens in a new tab)) Real customer and client work is multi-turn almost by definition. Someone asks a follow-up, changes their mind, or leaves out the detail that matters.

That is why many AI layoffs hit a wall. The tools are useful for single, well-defined steps. They are much weaker on the long, messy conversations that a service team, a support desk or an agency account team handles every day. I wrote about that gap from the governance side in what AI agents do at 3am versus in the demo (opens in a new tab).

Why does the rebound from AI layoffs matter to agencies?

Because the companies that cut are now short of exactly the skills an agency can supply. ManpowerGroup's 2026 survey of more than 39,000 employers in 41 countries found 72% report hiring difficulty, and for the first time AI skills top the list of hard-to-find capabilities: "AI Model & Application Development (20%) and AI Literacy (19%) now lead the global ranking." (ManpowerGroup (opens in a new tab))

IDC had warned about this earlier: "By 2026, 90% of organizations worldwide will feel the pain of the IT skills crisis, costing as much as $5.5 trillion in delays, quality issues, and revenue loss." (CIO (opens in a new tab))

Put those next to the rehiring forecasts. Companies made AI layoffs, found the work still needed doing, and now face a tight market for the people who can do it. An agency with real technical capacity is a faster answer than a hiring cycle.

The agencies that cut their own bench

Agencies were not immune to the thinking behind AI layoffs. Forrester expected agency headcount to fall 15% in 2026 after an average 8% cut in 2025. (Forrester (opens in a new tab)) Some of those cuts landed on developers and technical staff, on the theory that AI tools would cover the build work.

If that was your agency, the problem is timing. Your clients are asking for help rebuilding what their own AI layoffs removed, and you are competing for the same scarce people they are. A job post, a hiring loop and onboarding take months. The client wants an answer this quarter.

What does a resilient agency tech bench look like after AI layoffs?

The agencies I see doing well keep two things close and rent nothing they cannot replace. They keep strategy and the client relationship in-house. They get technical delivery from people who do that work every day, whether on staff or through a partner who builds under the agency's name.

That second model is white-label development. It is not shipping work to the lowest bidder. Done well, it is a build team that follows your standards, your contracts and your brand, and stays out of the room unless you invite it.

  1. Write down what you sell. List the technical deliverables in your last ten proposals: sites, integrations, automations, hosting. That list is your bench requirement.
  2. Mark what you can deliver today. For each item, note who builds it now. Any line with no name is a gap that AI layoffs elsewhere will push clients to ask about.
  3. Decide own or partner. Keep in-house what you sell constantly and what differentiates you. Partner for the rest, with written standards.
  4. Use AI where it is reliable. Single-step, well-defined tasks with a person checking the output. That is the lesson the AI layoffs taught the hard way. Quality gates catch the misses before a client does.
  5. Keep continuity. The same people on an account in year one and year three is worth more than any tool. Clients notice turnover faster than they notice features.

Questions to ask a build partner

  • Who owns the code and the data when the engagement ends?
  • Will my client ever see your name, and who decides that?
  • Where is the work hosted, and is that machine shared with other customers?
  • What happens when the AI part of a workflow is unsure? Is there a human path?
  • Can I see how you test and review work before it ships?

How should an agency talk to clients about AI layoffs?

Calmly, and with numbers. Many of your clients made AI layoffs or are under pressure to. Telling them they were wrong will not help. Showing them where the work went, and offering to pick up the parts that need people, will.

A simple frame works: what the AI handles reliably, what it hands to a person, and who that person is. If you can name the person and they are on your bench or your partner's, you are the answer to their rebound. If you cannot, fix that before the next renewal.

A short script for AI layoffs conversations

When a client brings up AI layoffs, at their company or a competitor's, three sentences cover most of it. First: here is the part of this work AI does well for you today, and here is how we check it. Second: here is the part that needs a person, and here is who that person is on our side. Third: if your own team got thinner, here is what we can pick up this month without a hiring cycle.

That script does two things. It treats AI layoffs as a business decision the client made for real reasons, without judging it. And it moves the conversation from the cost of people to the cost of the work not getting done, which is where agencies with a real bench win.

Frequently asked questions

Do companies regret AI layoffs?

Many do. Forrester found 55% of employers regret laying off staff because of AI, and it expects about half of AI-attributed layoffs to be reversed.

Are companies rehiring after AI layoffs?

Some already are, and analysts expect more. Gartner predicts half of companies that cut customer service staff due to AI will rehire by 2027 to perform similar functions, and Forrester expects much of the returning work to be paid less or moved offshore.

Why did Klarna bring back human customer service?

Its CEO told Bloomberg that cost had been too dominant a factor and the result was lower quality. Klarna now wants customers to always have the option to reach a person.

How good are AI agents at business tasks today?

Uneven. On the CRMArena-Pro benchmark, leading agents reached about 58% success on single-turn tasks and about 35% on multi-turn ones.

Should my agency rebuild its technical team after AI layoffs?

After AI layoffs, rebuild the capability, not necessarily the headcount. Keep strategy and relationships in-house and use a build partner for delivery you cannot staff quickly.

What is white-label development?

A partner builds the site, application or system and your agency delivers it under your own name, contracts and client relationship.

Where we fit

Kief Studio is a build partner for agencies that do the work, not agencies that sell the meeting. With white-label development, we build the site, application or system and your agency delivers it under your own name. We build to your brand, your contracts and your client's requirements, host and maintain it under your name when the engagement includes it, and stay out of the client relationship unless you bring us in. Our work runs on dedicated infrastructure, with no shared hosting, and we have 13+ year client relationships. When the job is AI automation, we build quality-gated systems with a human path for anything the model is unsure about, and run them for you on LTFI (opens in a new tab).

First conversation is free. No commitment.

Tell us what your clients are asking for

Related articles

  1. Build

    Utilization Rate Is Stuck Because Every New Service Line Still Needs a Body. Elastic Capacity Doesn't.

    Why your utilization rate looks low while you turn work away: hiring lag, queueing math, burnout, and how elastic partner capacity changes the staffing shape.

    Brian Gagne

  2. Build

    Should Your Agency White-Label a Core Service This Year? Run the Numbers First

    Should your agency white label a core service? What changed in 2026, where white label fails, and a one-sheet test to decide between a partner and a hire.

    Brian Gagne

  3. Build

    Agencies Are Cutting Jobs While Clients Still Need the Work. Agency Capacity That Flexes Is the Model That Holds.

    Agencies cut headcount 8% in 2025, with 15% forecast for 2026. Why hiring to grow is risky, and how flexible agency capacity lets a lean shop take on more work.

    Amelia Gagne