-1.png)
Tech's Layoff Reflex Is Backfiring. Here's What to Try Before You Cut Your Team
With AI layoffs already outpacing what companies expected to save and most leaders admitting the cuts didn't deliver, here's a fuller set of cost-saving moves worth trying.
In February 2025, Klarna's CEO Sebastian Siemiatkowski said he believed AI could already do all of the jobs humans do at the company. Klarna had claimed its AI assistant replaced roughly 700 customer service agents and froze hiring for a full year, building a public narrative around running “AI-first.”
Three months after that February statement, he was recruiting humans again. By May 2025, Klarna was posting freelance customer service roles, and Siemiatkowski put the reason plainly: “As cost unfortunately seems to have been a too predominant evaluation factor when organizing this, what you end up having is lower quality” (Fortune; Bloomberg).
Klarna isn't an outlier; IBM, Salesforce, Google, and Meta have all quietly rebuilt teams they trimmed in the name of AI efficiency, and the pattern shows up in the broader numbers too.
Only 8.4% of HR leaders say the AI-driven layoffs they carried out delivered what they promised, according to a Careerminds survey of 600 HR leaders published in July 2026. Roughly nine out of ten leaders behind this year's AI layoffs are, by their own account, not seeing what they signed up for.That number lands differently depending on where your company sits right now. Maybe the board hasn't said the word “layoffs” yet, but the budget conversations have gotten longer and more careful. Maybe a freeze is already in place, and you're trying to figure out what's still possible under it. Or maybe finance has already come down with a number, and the question in front of you is how to hit it without gutting the team that has to hit next quarter's roadmap too.
This piece isn't going to tell you layoffs are never the right call; sometimes a business genuinely shrinks, and a team built for a bigger scope needs to shrink with it. What it will do is walk through the moves worth trying first, matched to how much pressure your team is actually under, using real cost-saving results from Softjourn's own R&D work along the way.
The Numbers Behind the Reflex
Are Layoffs Increasing in 2026?
The numbers say yes. As of early September 2026, tech layoffs for the year had already reached 123,305 employees across 289 companies, edging past the 122,606 cut across all of 2025, with four months still left on the calendar. AI is a real part of that story – Nikkei Asia's reporting found that of the roughly 78,600 tech workers laid off between January and April 2026, nearly 48% of those cuts were attributed to reduced need for human workers because of AI and workflow automation.
Why Companies Are Already Regretting AI Layoffs
The same Careerminds survey found that among companies rehiring after AI layoffs, 35.6% brought back more than half the positions they'd eliminated, 32.9% reported losing critical skills they didn't plan to lose, and 75% said the layoffs hurt financial performance rather than helping it.
Put plainly: a large share of the companies making AI job layoffs this year are the same companies that will spend next year trying to undo them, and Klarna is simply the most public example of the pattern.
That reversal costs more than most budget conversations account for. SHRM's benchmarking research puts the average cost per hire at roughly $4,100 across all roles, and that figure mostly reflects internal recruiting time and job postings, not outside help. Bring in a recruiting agency to fill a technical role quickly, which is common when a company needs an engineer or DevOps hire back on short notice, and fees typically run 15% to 30% of that person's first-year salary. For a $150,000 engineering hire, that's $22,500 to $45,000 in fees alone, on top of the salary itself and whatever ramp-up time it takes someone to get back to full productivity (Dover).
We're not necessarily saying you shouldn't reduce headcount. But we'd argue it belongs at the end of the list of options, not the top of it, especially when the pressure driving the decision might be temporary, competitive anxiety rather than an actual funding or revenue problem.
Meet Yourself Where You Actually Are
Not every company reading this is in the same spot, and the right move depends heavily on which one describes you right now.
Stage | What it looks like | Where to start |
|---|---|---|
Watching | Budgets are intact, but leadership is asking “what’s our plan if this gets worse,” and competitors cutting staff is making the board nervous. | Audit spend that's already inefficient before touching people. Pilot AI-assisted workflows on one function so you have real numbers banked before you need them. |
Tightening | A hiring freeze or spend freeze is already in place. New full-time hires are off the table, but the roadmap hasn't moved. | This is the territory our Hiring Freeze Playbook covers in detail: what a freeze actually blocks, and three ways teams keep shipping without touching headcount. |
Under pressure | Finance has handed down a number, and headcount reduction is genuinely on the table as one option among several. | Our Staff Augmentation vs. Outsourcing vs. Layoffs piece walks through a framework for telling a temporary capacity gap from a permanent structural one, since that distinction should drive the decision. |
If you're in the first column, the good news is that the moves below are cheapest and most reversible when you make them early, before a freeze forces your hand. If you're already further along, the same moves still apply, they just carry more urgency.
Before adding or cutting a single person, it's worth asking a more basic question: how much more can the team already in seats actually produce? Softjourn's R&D group has spent the past year running structured experiments on that exact question, using AI tools on real project work rather than benchmarks, and documenting the results honestly, including where the tools fell short.
A few findings worth knowing about:
AI-Assisted Engineering: A 125% Increase in Delivery Throughput
One senior engineer built a fully autonomous, self-reviewing AI development loop on a live client engagement, where one AI agent writes the code and a second independently reviews it before a human ever looks at the output. Well-defined, clearly scoped tasks saw a 5 to 10x speed increase, and the client's own tracked metrics showed a 125% increase in delivery throughput, all while running near 100% of coding tasks through the workflow at roughly $120 a month in compute cost. Read More →
AI-Assisted Design: A 95% Faster Asset Rebranding Project
A large-scale rebranding project involving 150 themed assets across a complex codebase, the kind of work that normally takes a designer or developer 12 to 16 hours, was completed by an autonomous agent in 45 minutes, a 95% reduction in execution time. Separately, standard competitive UX research that usually takes several hours was completed in under an hour. Read More →
AI-Assisted QA: Cutting Sanity Testing Time by 3x
A proof of concept using low-code test automation cut a routine sanity test suite from 30 minutes to 10, a 3x reduction, and made that automation buildable by manual QA engineers without a dedicated automation specialist on staff. Worth noting: on a second project, the team tested the same tool and found it wasn't the right fit, and said so rather than forcing an adoption that didn't hold up. Read More →
None of this is a promise that AI will fully close a budget gap on its own, and Softjourn's own experiments have been candid about the tasks where it didn't work well – particularly anything requiring deep product-specific judgment. But as a first lever, before touching headcount, it's a considerably cheaper and more reversible one to pull.
The Case for Waiting to Layoff
Budget pressure creates a pull toward fast, visible decisions, and layoffs are the fastest, most visible one available. But not every decision under pressure needs to be made immediately, and most of the genuine alternatives to layoffs or downsizing aren't secret tactics. They're really just a matter of sequencing: try the reversible moves before the irreversible one. Some ways to postpone layoffs are:
Delay the Big Infrastructure Decisions
Delaying a full re-platform or major tooling consolidation until you have real usage data, rather than making the call during the same quarter the pressure hit, tends to produce a better decision and a cheaper one. Panic-driven infrastructure changes are hard to reverse and easy to regret.
Bring in Outside Engineering Support for Defined Projects
Moving a defined, secondary project to an outside team instead of pausing it indefinitely or pulling internal engineers off their current priorities keeps the work moving without expanding payroll or disrupting the roadmap your main team owns. This is a different move than staff augmentation. It hands a whole piece of work to a partner that owns delivery end to end, which tends to fit projects with a clear scope and a defined finish line, like a platform build-out or a discrete feature project.
Give Temporary Pressure 90 Days Before Cutting Staff
Perhaps the hardest one: giving a temporary capacity gap 90 days to prove it's actually temporary, rather than reaching for a permanent headcount cut in week one. Given that 35.6% of companies surveyed ended up rehiring for more than half the roles they eliminated, and that it took Klarna just three months to go from “AI can do all of our jobs” to recruiting humans again, that waiting period would have been considerably cheaper than the round trip most of them ended up taking.
Waiting isn't the same as doing nothing. It means using the moves above – the audits, the AI-assisted workflows, the outsourced project work – while giving yourself the room to find out whether the pressure is a season or a permanent shift before making a call that's expensive to reverse.
Before You Reach for Headcount
The data from this year makes a case that's easy to miss in the moment: the tech industry is currently proving, in public and at scale, that the fastest cost-saving move isn't always the cheapest one once you count what comes after it. Klarna's own CEO said as much when cost turned out to be too predominant a factor in a decision that hurt quality. A large share of the companies making AI layoffs this year will spend part of next year rebuilding what they cut.
None of that means your situation is the same as theirs, or that a real, structural reduction is never the right call. Our point: auditing spend, testing AI-assisted workflows on real work, outsourcing defined projects instead of pausing them, and giving temporary pressure all deserve the same seriousness as a layoff decision. And best of all, they're considerably easier to walk back if they don't pan out.
If your team is trying to figure out where it sits on that spectrum, or wants real numbers on what AI-assisted development could add to your current capacity, contact Softjourn to talk through what a plan built for your specific situation could look like.
