Between Anthropic, Google, and GitLab, the biggest names in AI are racing to replace the tools your business pays monthly for. What the shift means and whether SMEs should follow the lead.
What’s happening: AI agents, software that can autonomously complete multi-step tasks without human input, are disrupting the foundation of the global software industry. In early 2026, roughly $2 trillion in SaaS market value was erased as investors bet the per-seat subscription model is becoming obsolete.
Why this matters: Many SMEs are already paying for tools they barely use, the rise of affordable AI agents raises a direct question: is it time to stop buying seats and start building smarter?
Something significant shifted in the global software industry in early 2026, and most Australian small business owners have not heard about it yet.
Between January and February, roughly $2 trillion in market capitalisation was erased from software companies around the world. The trigger was not a recession or a regulatory crackdown. It was AI agents, autonomous software systems capable of completing entire workflows that previously required a team of people and a stack of paid tools to manage.
The term “SaaS-pocalypse” emerged almost immediately to describe what was happening. Analysts and investors began asking a simple but uncomfortable question: if AI agents can perform many of the tasks that SaaS tools were built to support, what happens to the business model that charges per user, per month, indefinitely?
The latest move in this space came this month, when GitLab announced an expanded collaboration with Google Cloud, allowing AI agents in its development platform to call Google’s Vertex AI models directly. Days later, Anthropic launched Claude Managed Agents, a service designed to let organisations run autonomous AI agents without managing the underlying infrastructure themselves. Early adopters including Notion, Asana, and Sentry have already integrated the service for tasks spanning code automation, HR, and finance workflows.
The seat problem
To understand why this matters for small businesses, you need to understand how most software pricing works. For two decades, SaaS companies built their revenue models around seats. Every time a business hired someone, that person needed a login. More staff meant more seats, which meant more monthly revenue for the software provider.
AI agents break that correlation entirely. A single agent can perform the work previously done by multiple logins across multiple tools, and it does not need its own seat.
The consequences of that shift are already visible. One investor cited in a TechCrunch analysis of the SaaS-pocalypse noted that the barriers to building custom software have fallen so dramatically that “the build versus buy decision is shifting toward build in so many cases.” A separate analysis described marketing agencies that once required teams of ten using complex software stacks now achieving similar output with two people and a set of autonomous agents.
For Australian SMEs, the numbers behind this shift are significant. According to Gartner’s Australian public cloud forecast, local businesses spent close to $13 billion on SaaS alone in 2025, a 15.5% increase on the prior year. Research cited by Sydney-based consultancy Scale Suite suggests that between 25 and 30% of SaaS licences are unused or significantly underutilised. Applied to a $50,000 annual software budget, that translates to between $12,500 and $15,000 in wasted spend each year.
The most widely cited benchmark for SME technology spending sits between 2 and 7% of annual revenue. For a business turning over $2 million, that is up to $140,000 a year. A meaningful portion of that is going to per-seat subscriptions that an AI agent could, in many cases, replace today.
Where Anthropic fits in
Anthropic, the San Francisco-based AI company backed by Google and Amazon, has moved aggressively into the business automation space in 2026. Its Claude Managed Agents service, launched in April, removes the infrastructure overhead from running AI agents, allowing businesses to deploy autonomous workflows without engineering teams or server management.
Claude Code, Anthropic’s AI coding tool, reached $1 billion in annualised revenue within six months of its public release in mid-2025. The company now captures a majority share of spending among businesses buying AI tools for the first time.
The practical implication for small businesses is that access to capable AI agents is no longer the exclusive domain of large enterprises. A subscription to Claude Pro costs $20 to $50 a month. That is a fraction of what most businesses pay for a single seat in a dedicated project management, customer service, or analytics platform, let alone several of them running simultaneously.
Forrester Research has noted that SaaS vendors offering genuinely differentiated or highly specialised solutions, particularly in regulated industries like healthcare and finance, are more likely to hold their ground. The tools at greater risk are the generalised ones: generic project trackers, basic CRM systems, templated reporting dashboards, and workflow tools that an AI agent can now replicate without a monthly invoice.
What SMEs should do now
None of this means small business owners should cancel all their software subscriptions tomorrow. The transition to agentic AI is real, but it is not instant, and the tools that survive will be those embedded deeply enough in a business’s data and processes that replacing them carries genuine risk.
What it does mean is that the annual software renewal is no longer a formality. Every subscription worth reviewing should be reviewed against a simple question: is this tool doing something an AI agent cannot, or am I paying for a seat that a $20 monthly subscription could replace?
A 2026 survey by Databricks found that the use of multi-agent systems grew by 327% over a four-month period, with 78% of companies now using at least two AI model families in their operations. That adoption curve is moving faster than most small businesses realise.
The businesses best positioned to benefit are those already comfortable experimenting with AI tools, those with tight margins who feel the weight of per-seat costs most acutely, and those operating in states like Queensland and Western Australia where economic activity and credit demand are running strongest right now. The SaaS-pocalypse is not a story about software companies losing money and business owners paying for tools they barely use, it might just be the best news they have heard all year.
Keep up to date with our stories on LinkedIn, Twitter, Facebook and Instagram.
