Every few years, the tech industry and its investors convince themselves an entire category is about to die. This year, it’s software-as-a-service, or SaaS.

The rampant view goes that if AI can generate code, software becomes cheap and software companies lose value. Cue the sell-off, the doom loop and stories featuring analysts predicting the “end of software”.

It’s a seductively simple theory, but it’s both incomplete and indiscriminate. It fails to distinguish between software that looks useful and software that can be trusted to run something that actually matters.

AI is making software far easier to build, with small teams now able to produce tools in days that once took months. But this new abundance does not erase value, it changes where value sits.

If you have been around as long as I have, you have seen this before, and more than once. When video content became cheap, value did not disappear from media. It moved to distribution and curation. When retail moved online, value did not disappear from commerce. It moved to logistics, fulfilment and customer loyalty. The same pattern is playing out now with SaaS.

In this new world, fragile, shallow tools will struggle. Systems that hold trusted data, enforce rules, connect knotty workflows and survive real-world scrutiny will endure.

Much of the SaaSpocalypse commentary treats software as a single category, differing only by interface and development speed. But there is a material difference between a tool that automates a narrow task and a system that sits beneath payroll, payments, inventory, tax or regulatory reporting. The former is a feature. The latter is institutional memory.

No one with a brain is going to vibe-code a system of record such as a finance or payroll platform. Such a system has to deal with exceptions, permissions, compliance rules, audit trails, version control, data integrity, security reviews and the awkward reality of how organisations actually operate.

None of that goes away with better tooling. If anything, it becomes more important because, as AI systems move from making suggestions to taking action, the cost of bad software rises. A consumer app that works most of the time is irritating, but a finance or compliance system that works most of the time is a matter of existential liability.

This is where many investors are misreading the moment. They see the collapse of the interface moat and assume the entire castle is gone. But interfaces were rarely the strongest defence and, in many categories, they were the weakest.

The durable advantages sit elsewhere: proprietary data, embedded workflows, ecosystem relationships, regulatory understanding and implementation capability.

Think about data. The most valuable data in a business is not scraped from the internet. It’s the operational truth generated by running the business: what was sold, what was paid, what failed, what passed compliance, which customers stayed and which suppliers slipped.

That data compounds over time. It’s difficult to recreate and hard to migrate easily. Paired with AI, it becomes even more valuable. It becomes the difference between a system that sounds intelligent and one that behaves intelligently.

Then there’s integration. On paper, APIs make everything connectable. In reality, most organisations are still a dizzying patchwork of systems, spreadsheets and workarounds built up over years. The software that matters can sit inside that complexity and make it coherent.

That work is hardly glamorous, but it’s defensible. Businesses don’t rip out core systems lightly because they are tightly integrated into dozens of processes that are risky to disrupt.

Regulation is another blind spot. In regulated sectors, accuracy is not a feature. It is literally the product. Software must be auditable, secure, explainable and aligned with rules that change slowly and carry real consequences.

That kind of capability is not protected by code scarcity but built through accumulated domain knowledge and institutional trust.

So, what we are seeing with SaaS now is more of a sorting than an extinction.

Yes, there will be a flood of new small applications and many will be useful, even excellent. But plenty will be short-lived and disposable. Categories built on thin differentiation and per-seat licensing will come under pressure as AI absorbs parts of their value.

But software itself won’t disappear. It is being unbundled and re-layered.

At the top, there will be an explosion of lightweight tools, agents and micro-applications. In the middle, orchestration and interaction layers. At the bottom, systems that store truth, enforce constraints and connect to the real economy.

Value is moving down the stack. That shift has implications for Australia. Our technology conversation is often pulled towards novelty and imported hype. But durable local winners are unlikely to come from building another chatbot wrapper or a marginally better dashboard.

They will emerge from businesses that understand complex local workflows, compliance-heavy sectors and data-rich operational environments.

Places where trust is hard won and failure is expensive. Places where buyers care less about demos and more about whether the product survives procurement and works on Monday morning.

By Charlie Wood, CEO Wiise

This article was first published by Capital Brief