# The SaaSpocalypse is a people problem > Source: > Published: 2026-08-26 13:00:00+00:00 There is a tidy story going around about the end of enterprise software. Call it the SaaSpocalypse. AI and vibe coding have made it cheap enough to rip out your software-as-a-service contracts and build your own replacements. The rush to rebuild carries its own risk, one that surfaces only after the SaaS is gone. Teams can almost always build the replacement. What they build, too often[, is a copy of what they already had.](https://www.cio.com/article/3847331/operational-efficiency-is-not-transformation.html) Few people are better placed to see that risk than Mike Anderson. As chief digital and information officer at Netskope, the cloud security company that went public on the Nasdaq in September 2025, Anderson runs both IT and the company’s strategy office, a seat that spans his own operations and the broader go-to-market. He is a 2026 inductee into the CIO Hall of Fame, sits on a long list of advisory boards and venture capital innovator networks, and is one of the industry’s most connected executives, fielding peer questions about AI nearly every week. Before Netskope, he was CIO for North America at Schneider Electric. He lives in Dallas. It starts with how fast the ground has moved. “We’ve gone from AI is my assistant, to I’m delegating a task to an agent, to I’m actually delegating full bodies of work to agents,” Anderson says. “We’re in that last one now.” The trouble is that our instincts have not caught up. Anderson is not interested in slowing anyone down. “I’m a big believer in innovation at the edge of your company. Innovate closest to the people, closest to the problem,” he says. “As CIOs, we don’t want to be the people who say no. We want to let them experiment and learn.” The danger he points to is quieter than recklessness. It is the pull to aim powerful new tools at rebuilding the past. “The risk is we’re not thinking differently. We’re thinking based on the bias of how things work today,” he says. “Today’s systems are built around people: dashboards that serve us insights, forms we fill in, workflows that pass work between teams. If we go vibe code something, it’s probably going to look a lot like that,” Anderson says. “And it’s not designed for agents, who don’t need a form and don’t need the dashboard. They just need access to the data, the API to call or the other agent to talk to.” As Anderson sees it, the opportunity is bigger than software. “It’s about reinventing processes with agents in the middle of the process,” he says. Doing that well means fighting the urge to start building. “Before you put any fingers on keyboards, get a small cross-functional team together, [look at reimagining the process](https://www.cio.com/article/3962000/when-transformation-needs-a-home.html) and start from the outcome. Then work back,” Anderson says. It also means changing the question. The reflex has been to ask whether a task can be done with AI. Anderson wants a sharper test. “It’s this work I could delegate to an agent in a [deterministic way, where there’s predictability in the outcome](https://www.cio.com/article/4184020/playing-to-win-at-the-ai-casino.html),” he says. “That’s a different pivot from where we were three or six months ago.” The teams that pull this off do not need to be big. Anderson keeps them deliberately small: a subject matter expert who lives the problem, a [product owner](https://www.metisstrategy.com/product-management-ai-value-measurement/) who frames the context and an engineer who turns it into something an agent can act on. “The old rule was a two-pizza team,” he says. “Now maybe the two pizzas are for three people who are just really hungry, because they’re working tirelessly.” Before any of that, Anderson puts discipline around what gets built at all. [Every candidate is weighed against three levers](https://www.cio.com/article/4112436/how-oshkosh-corp-turns-ai-hype-into-measured-business-value.html): whether it accelerates growth, whether it takes out cost and creates leverage, and what the risk is if it goes wrong. Then come what he calls the primitives: consistent user management, observability in the tools and standards encoded where agents will read them. “I have markdown files that determine the technologies I want used, down to the database,” he says. “I don’t want agents deciding today that they’re going to introduce a database that’s never been in my environment, because at some point this has to move to a production state.” That last part is what he thinks teams underestimate. When you replace a vendor, you inherit the job the vendor used to do. “Someone has to keep it running. We didn’t have that responsibility in SaaS. Now we do,” he says. The foundations reach past code, too: style guides, shared libraries, reusable AI assistants and skills that help non-engineers describe what they want, and documentation he insists must be “a first-class citizen.” Netskope IT team built one called Professor Vibe Code that turns a recorded description of an outcome into instructions an AI can build from. “It’s not ‘I need a field on a screen,'” he says. “It’s describing what you want as context with a clear definition of success.” For all the talk of architecture, Anderson keeps [steering back to people](https://www.cio.com/article/4158552/scaling-ai-at-union-pacific-starts-with-people.html). “We’ve always said building the technology is easy. Getting people to use it is hard,” he says. “That’s even more true here.” He has lived through the internet, SaaS and cloud, and ranks none of them with this. “I can’t point to a technology that’s as disruptive, or that introduces as much change, as AI.” Which is why he now treats his chief human resources officer as just as critical a partner as his CISO. The two of them talk constantly about the human risk of AI and how to bring people along. “Everyone is worried. Is AI going to replace me? What’s my future in an AI-first world,” he says. “So much of this comes down to giving people clarity about where they fit.” Borrowing from Maslow, he notes that psychological safety rests on more basic needs, the paycheck and the roof, and [that people](https://www.cio.com/article/4074675/the-clear-advantage-of-an-80-20-ai-operating-model.html) who feel threatened do not stay neutral. “Without psychological safety, on one extreme you get AI sabotage. On the softer end you get passive resistance, where people just resist using the AI,” he says. The aim is to keep humans at the center of processes that increasingly run without them. For CIOs facing the same moment, Anderson’s advice comes down to three moves. Start with a phone call. “If you’re not talking to your CHRO, get them on speed dial,” he says. “We’ve always kept the CISO on speed dial because security matters so much. You have to add the CHRO now.” Then be honest about the impact, even when the picture is incomplete. “You may not have all the answers today. Telling people that is important,” he says. And lay the foundations so teams can build at the edge without recreating the old pattern of work handed off later with no context. The ground keeps shifting, and Anderson does not pretend otherwise. The leaders who come out ahead, in his telling, will be the ones who [redesign the work,](https://www.cio.com/article/4080069/inside-the-product-mindset-that-runs-7-eleven.html) lay the foundations and never lose sight of the people doing it. For him, the discomfort is the job now. “The world is moving at a pace I’ve never seen before,” he says. “Every day, I have to get comfortable being uncomfortable.”