The Cracks in Traditional SaaS
Microsoft's CEO Satya Nadella recently made a striking prediction: business applications as we know them will "collapse" in the agent era. His argument is that AI agents will handle business logic across multiple systems, making the traditional software backend—the thing you're actually paying for with most SaaS subscriptions—increasingly obsolete.
This isn't just speculation from a tech executive. Venture capital is pouring into "vertical AI" companies—businesses that use AI to perform tasks previously done by humans, bundled as a service. The thesis is that these vertical AI companies could eventually dwarf traditional SaaS in value.
Meanwhile, the cost of building software is plummeting. Tools that let developers build functional applications in days for almost nothing are proliferating. The traditional value proposition of software—that it's expensive and hard to build, so you should rent access to someone else's—is being fundamentally questioned.
What This Means for Your Business
If you're a business that relies on SaaS tools, this shift matters. Here's what's changing:
Generic software becomes less valuable. If AI can orchestrate actions across multiple systems, the value of any individual system diminishes. The competitive advantage shifts from "having good software" to "having the right data and processes for AI to work with."
Custom becomes more accessible. As development costs fall, the calculus around build-versus-buy changes. Software that precisely fits your business—previously a luxury for large enterprises—is increasingly within reach.
Integration matters more than features. When AI agents need to work across your systems, how well those systems talk to each other becomes critical. Proprietary, siloed SaaS products become liabilities rather than assets.
Your data is your moat. As software commoditises, the businesses that thrive are those with rich, well-structured data that AI can actually use. Generic SaaS often traps your data in formats optimised for the vendor, not for you.
The New Landscape
SaaS providers are responding to this shift in various ways. Some are adding AI features to existing products—sometimes genuinely useful, sometimes marketing theatre. Others are building AI assistants that help users navigate their software. A few are going fully agentic, embedding autonomous AI directly into their platforms.
But the most interesting development is the rise of "service-as-software"—AI systems that don't just help you do work, but actually do the work themselves. Instead of buying software for your team to use, you buy AI that performs the function directly.
This inverts the traditional model. Rather than paying for tools and hiring people to use them, you pay for outcomes and let AI figure out the tooling.
What Smart Businesses Are Doing
The businesses that are navigating this transition well share some common characteristics:
- They're questioning the SaaS patchwork. That collection of fifteen different subscriptions, each doing one thing adequately? It's worth asking whether a unified, custom system might now be both better and more cost-effective.
- They're investing in their data. Clean, well-structured, accessible data is the foundation for everything AI can do. Businesses are consolidating their data and making sure it's in formats they control.
- They're building for integration. New software decisions prioritise API-friendliness and interoperability. Proprietary lock-in is increasingly seen as a strategic risk.
- They're considering custom where it matters. For core business processes—the things that differentiate them from competitors—custom software is back on the table.
Why This Favours Custom Development
In the old world, SaaS made sense because software was expensive to build and maintain. You rented access to someone else's investment.
In the new world, that equation has changed. AI-assisted development has dramatically reduced the cost of building custom software. The ongoing maintenance burden—long the hidden cost of custom development—is also shrinking as AI makes code easier to understand and modify.
At the same time, the limitations of generic SaaS are becoming more apparent. Your business processes are unique. Your competitive advantages are specific. Software that forces you into someone else's workflow is increasingly a handicap, not a shortcut.
This doesn't mean every business should rush to build custom software for everything. Generic tools still make sense for generic needs. But for the core systems that run your business—the ones where your specific processes are your advantage—the case for custom has never been stronger.
The Path Forward
We're still early in this transition. Traditional SaaS isn't going to disappear overnight—there's enormous inertia in enterprise software adoption, and many SaaS businesses will successfully adapt.
But the direction is clear. Software is being commoditised. AI is changing what "using software" even means. And businesses that cling to the old model—renting generic tools and adapting their processes to fit—will increasingly find themselves at a disadvantage.
The winners will be businesses that own their core systems, control their data, and build software around how they actually work rather than how a vendor decided everyone should work.
That's a significant shift. And for businesses ready to make it, models like CoSaaS offer a way to get custom software without the traditional upfront investment and risk.
If you're rethinking your software strategy in light of these changes, we'd be happy to talk through the options.