Digital Transformation Used To Mean Better Tools. Now It Means Fewer People
Konstantin Klyagin is the Founder of Redwerk and QAwerk, driving innovation in custom software development and quality assurance since 2005.
gettyI’ve been running a software agency for two decades. For years, when clients requested “digital transformation,” I knew what they meant. Move some things to the cloud. Clean up the reporting. Kill a few spreadsheets. Give people dashboards so they’d stop asking each other for updates. It was modernization with a nicer name.
However, something has shifted. Even the buzzwords are changing. Rather than just “digital transformation,” I increasingly hear “AI workforce transformation.” Instead of better tools, businesses want to know how much of the work AI can take over.
A client walked into a conversation with us not long ago whose personnel bill runs around $2 million a year. He didn’t want a portal. He didn’t want another dashboard. He wanted to look at every workflow in his business and figure out how much of it should still be done by humans.
Executives are starting to question the very structure of their daily operations. A CFO I spoke with walked me through her approvals workflow and pointed at a $400 purchase that had needed five signatures. She wanted to know whether four of them served any real purpose or were just a habit nobody had questioned. That kind of question used to come from our side, during discovery. Now clients bring it to the very first meeting.
Every company I’ve worked with has at least one internal system that actually manufactures extra jobs. For example, someone reconciles exports, someone reformats reports, someone chases sign-offs the tool was supposed to eliminate. Previously, we’d simply swap an outdated tool for a shinier one or build custom software for the same process. Now, companies are questioning if the entire process should even exist.
To be fair, even in the old days of custom software, we always tried to rethink how work got done and eliminate unnecessary steps. But today, the focus on cutting out useless tasks is much stronger, driven by a real push to shrink workforce expenses.
Recently, big tech has announced successive layoffs, often crediting AI for increased efficiency. Smaller firms aren’t trying to imitate the layoffs, but they’re absorbing the message: Maybe we’re staffed for an operating model that’s already obsolete.
A founder I know sat down recently with a spreadsheet showing every person in his company next to a rough estimate of how much of their week could, in theory, be done by software. He’s the fifth person to describe that exercise to me this quarter, which is what convinces me the shift is real. However, wrong cuts at smaller companies are disastrous; you might accidentally lose the person silently managing three workflows from memory, only noticing when a customer complains.
That’s why the $2M payroll conversation felt more mature than most. The client was asking for a redesign, which is another thing altogether.
A client of mine cut three roles in accounts payable last year, citing an AI invoice-processing rollout. Real savings were a few hundred thousand a year. Half a year later, they hired someone new full-time. Her job was to catch the invoices where the model had confidently “read” a wrong vendor number or duplicated a line item. The layoff and the hire didn’t cancel out on paper, but they came pretty close.
Bad automation actually relocates labor, usually to more expensive and more frustrated people. It’s what happens when a company rushes to be “AI-first” before anyone’s clarified what it’s actually automating away.
Take Klarna, for example. They announced their AI customer-service agent had replaced the work of 700 human reps, but started re-assigning marketers and engineers to customer support a year later when quality dropped. We’ve seen similar stumbles elsewhere, like McDonald’s pulling the plug on its AI drive-thru partnership with IBM after customer orders were continually botched, forcing human cashiers to step back in.
What happened at these massive corporations is what I keep watching in miniature at smaller companies. The AI handles its impressive bit, but something the model can’t quite handle piles up in the background. And by the time somebody notices, the company is already rebuilding the human layer, often at higher cost, because the original people have moved on.
Through our agency’s QA testing, we get a strange vantage point on automation. You get to see what happens a couple of release cycles after the launch celebration: The tickets pile up because the model was confidently wrong about the 10% of cases nobody stress-tested. Handling those tickets is a real job, but it never appears on any of the slides that got the automation funded.
This brings me to the part I find genuinely interesting. The competitive divide is no longer old tools versus new, as almost everyone has new tools now. It’s between companies that are merely digitalized and companies that have actually redesigned how work flows through them. One group bought software; the other used it as an excuse to ask harder questions.
This current wave of digital transformation is basically a deep restructuring of the company, and AI is just the most effective tool for the job right now. It can do remarkable things in careful hands and terrible things in careless ones. The startups I see surviving the current SaaS shakeout are the ones that thought hardest about what still deserves a human.
If you’re a smaller enterprise watching the layoff headlines and wondering whether it’s your turn to “get lean,” resist the urge to lead with cuts. Lead by mapping out your actual operations and understanding what your people actually do, why they do it and which of those reasons still hold. The savings will come from work you decided not to do, not from people you decided not to keep. That’s a much better story to tell your team, your customers and yourself.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
