By Adam Mills
AI Editor
0
Share
Newsweek is a Trust Project member
See more of our trusted coverage when you search.Prefer Newsweek on Googleto see more of our trusted coverage when you search.
Inside many enterprises, the build-versus-buy debate is no longer theoretical. Teams are quietly replacing specific workflows with tools they create themselves—sometimes in days, sometimes without IT oversight and often before leadership has fully adjusted its governance models.
A new report from Retool, a software platform that helps companies build internal tools and workflows, suggests that the move from buying software to building it internally is already underway. The findings draw on a late-2025 survey of 817 Retool customers and builders.
Participants represented roles across engineering, operations, product management, data, IT, marketing, business analysis and finance at companies ranging from startups to Fortune 500 enterprises.
Read More on Newsweek.AI
Thirty-five percent of respondents said they have replaced functionality of at least one SaaS tool with a custom build, and 78 percent expect to build more of their own tools in 2026. At the same time, 60 percent reported building something outside of IT oversight in the past year.
The implications extend beyond simply replacing vendors. As software building becomes more accessible across business units, the question for executives is no longer whether teams will build. It is how that building is structured, supported and monitored.
The Economics Behind the Shift
David Hsu, CEO of Retool, says that the economics of internal software have fundamentally changed.
“The cost of building custom software has collapsed,” Hsu told Newsweek. “Two years ago, a custom internal tool might take an engineering team weeks and cost six figures. Today, a business operations lead with the right platform can have a working prototype in a day or two.”
“That's a structural change, not a cyclical one,” he said. “When the cost of building drops by an order of magnitude but the cost of buying stays flat, the math changes for every company.”
...
In practice, companies are not abandoning core enterprise systems altogether. Instead, they are targeting areas of friction—approval flows, dashboards, administrative panels—that never fully matched their internal processes.
“It’s almost always narrow workflows first,” Hsu said. “Nobody's ripping out Salesforce wholesale. What they're doing is replacing the specific piece of a tool that never quite fit.”
Those incremental replacements can compound.
“The default question shifts from ‘What should we buy?’ to ‘Can we build this?’” Hsu said.
Why Governance Is the Real Fault Line
The report’s most revealing data point may not be the replacement rate, but where the building is happening. Sixty percent of builders reported creating tools outside of IT oversight in the past year.
“Shadow IT at this scale is a demand signal,” Hsu said.
Procurement processes were largely built, Hsu noted, for an era when software development was harder and more expensive than it is today.
But building a prototype is not the same as deploying production software. The report found that 51 percent of builders have shipped production tools currently in use by their teams. Moving from prototype to widespread deployment depends less on the AI model and more on the surrounding systems, Hsu argues.
“The dividing line is almost never the AI, it's everything around the AI,” he said. “A prototype that runs on sample data is impressive. A production tool that connects to your actual Salesforce instance, respects role-based access, and passes a security review is useful.”
That distinction between experimentation and infrastructure mirrors a broader pattern playing out across enterprise AI adoption.
Suraj Srinivasan, professor at Harvard Business School, recently told Newsweek that organizations face an implementation gap between AI’s capabilities and what they are prepared to deploy. Companies are often cautious about privacy, customer harm and safety risks, he said, which can limit how fully they use the technology.
“But that means you cannot get the most value from the technology,” Srinivasan said.
In practice, that caution often pushes companies toward lower-risk, back-office deployments while more transformative applications remain out of reach. The same tension now applies to internal software building: Capability may be expanding quickly, but the surrounding governance, security and review processes must mature alongside it.
Hsu says the bigger concerns emerge once tools are live.
“The most underestimated risk is what happens after deployment,” he said. “The real risk isn’t the individual tool, but the ungoverned sprawl.”
A single internal tool may introduce limited exposure. Dozens of unsanctioned applications connected to production data, and maintained by unclear owners, can create a growing attack surface and compliance burden.
Measurement as Infrastructure
Governance is only part of the maturity equation. Thirty-seven percent of organizations surveyed by Retool said they have not yet established AI productivity metrics.
For Hsu, that gap threatens to undermine the shift.
“You can't prove ROI on what you don't measure,” he said.
SaaS subscriptions, he noted, came with built-in financial visibility. Custom-built tools require deliberate tracking of time saved, processes eliminated and downstream cost reductions.
“Experiment, but measure the experiments,” he said.
The study shows that about half of builders who have shipped production software report saving six or more hours per week. That time savings provides a starting point, but leaders must go further, linking those gains to decision quality, operational resilience and risk reduction.
Only 19 percent of respondents described their organizations as having advanced AI automation maturity. According to Hsu, the difference is not simply more AI usage, but stronger guardrails.
“Advanced maturity means building is the default, not the exception,” he said. “It happens inside a governed environment with full visibility.”
The Future of SaaS
The rise of internal building does not signal the end of purchased software, but it does raise expectations.
“The bar for purchased software just got permanently higher,” Hsu said.
Rather than wholesale replacement, he expects an evolution.
“What I don't think happens is a wholesale collapse of SaaS,” Hsu said. “There will always be categories where buying makes more sense than building. But the number of categories where that's true is shrinking.”
For enterprise leaders heading into planning cycles, the strategic decision is less about ideology and more about control.
“The most important decision isn't whether to build—your people are already building,” Hsu said. “The decision is whether that building happens in a governed environment or in the shadows.”
Attempting to suppress internal building may push it further underground. Embracing it without structure may invite fragmentation.
“The winning move is to give builders a platform where they can move fast and build securely, connected to real data, with permissions and audit trails from day one,” he said. “Channel the energy, don't fight it.”
As the economics of internal software continue to shift, enterprises may find that the build-versus-buy debate was never the central question. The real dividing line is between organizations that treat governance as foundational infrastructure and those that discover the consequences only after the sprawl has already taken root.
Newsweek will explore these governance challenges further during its upcoming AI Governance webinar on Feb. 24, featuring Suraj Srinivasan, professor at Harvard Business School and Keith Enright, partner at Gibson Dunn, co-chair of the firm’s Tech and Innovation Industry Group and Artificial Intelligence Practice Group, and former chief privacy officer at Google.
Register for the webinar, for free, today.
View 0 Comments
Submit Correction
View Editorial & AI Guidelines