Category: Technology | Published: 2026-05-19
The Software Industry Is Changing Faster Than Most People Realise
For the past two decades, enterprise software has run on a remarkably consistent model. Businesses buy licences based on how many employees need access, vendors charge per seat, and revenue grows as headcount grows. It is a model that turned companies like Salesforce, Microsoft, Slack, and Workday into some of the most valuable businesses on the planet.
That model is now under serious pressure, and the cause is artificial intelligence.
AI agents are not just another feature added to existing software. In a growing number of cases, they are becoming the primary worker inside a system, handling customer queries, generating content, analysing documents, processing requests, and automating workflows with limited human involvement. When the software itself starts doing the work that employees used to do, the logic of paying per employee begins to fall apart.
The transition is still early, but the numbers suggest it is already well underway.
What The Spending Data Actually Shows
Procurement platform Tropic analysed more than 18 billion dollars in managed software spending across mid-market and enterprise organisations. The results were striking. AI-native enterprise software spending grew by approximately 94 per cent year-on-year, while spending on primarily traditional software platforms grew by around eight per cent over the same period.
That is not a marginal difference. It represents a fundamental shift in where business technology budgets are flowing, and it is happening quickly enough that most established software vendors are scrambling to respond.
Research from Deloitte supports the same direction, suggesting that software companies are under mounting pressure to become AI-first businesses. Agentic AI, where software operates autonomously rather than simply assisting users, is expected to transform pricing models, software operations, and customer expectations across the industry. Gartner goes further, predicting that by 2030 at least 40 per cent of enterprise software spending could shift from traditional per-seat licensing towards usage-based, agent-based, or outcome-based pricing.
These are significant predictions from credible sources, and they reflect a genuine structural change rather than typical technology hype.
The Difference Between AI-Assisted And AI-Native
Not all AI in software is the same, and the distinction matters for businesses trying to understand what they are actually buying.
Traditional software platforms are built around human users operating interfaces manually. Adding AI features to those platforms, such as Copilot inside Microsoft 365 or Einstein inside Salesforce, produces a hybrid model where AI assists people but humans remain the primary workers. This is the category most large software vendors are currently in.
AI-native platforms are built differently from the ground up. In these systems, the AI itself is the main worker, and humans step in to review, direct, or correct rather than to perform tasks. Some newer customer service platforms, for example, now allow businesses to deploy autonomous AI agents across email, live chat, social media, and messaging apps simultaneously, with a human team managing exceptions rather than handling every interaction directly.
The same pattern is emerging in areas such as software development, contract analysis, financial reporting, internal helpdesks, and operational workflow management. The AI does the work; people govern the outcome.
This is what makes AI-native different in kind rather than just in degree, and it is why the economics of the software market are starting to shift so significantly.
Why Established Software Vendors Are Moving Fast
The pressure on traditional software firms is not abstract. If an AI agent can perform the work that previously required five employees each holding a separate software licence, then a major pillar of the per-user revenue model disappears.
This is why companies like Salesforce, Microsoft, Google, ServiceNow, and OpenAI are all investing heavily in agentic AI products and promoting autonomous workflow capabilities as central to their future. They are not doing this purely out of enthusiasm for new technology. They are responding to a genuine threat to their existing revenue structures from AI-native competitors that are building from scratch without the legacy of per-seat models to defend.
The result is an industry in active transition, with pricing models, product architectures, and competitive dynamics all shifting at the same time.
For businesses buying software, this creates both opportunity and complexity. On the opportunity side, AI-native systems genuinely can reduce manual workload, improve response times, and automate repetitive tasks that currently consume significant staff time. On the complexity side, the move away from predictable per-seat pricing towards models based on usage, AI actions, compute consumption, or outcomes makes budgeting and forecasting considerably harder.
The Hidden Cost Risk Businesses Should Know About
One aspect of the AI software transition that tends to be underestimated is the cost structure of running AI systems at scale.
Unlike traditional software subscriptions, which have fixed and predictable monthly or annual costs, AI systems consume compute resources, API calls, processing tokens, and cloud infrastructure in ways that vary significantly depending on usage. When an AI agent handles a thousand customer queries, it consumes resources in a way that a human using a CRM does not.
Research cited by Tropic suggests many organisations are already experiencing AI-related software price increases well above normal annual SaaS uplifts. Businesses that adopt AI-native platforms without carefully modelling usage-based costs can find themselves facing bills that are considerably higher than expected.
This does not mean AI software is not worth it. In many cases the efficiency gains more than offset the costs. But it does mean that procurement decisions in this space require a different kind of analysis than simply comparing per-seat licence prices.
What This Means For UK Businesses Right Now
The AI-driven transformation of enterprise software is not something that only affects large corporations or technology companies. It is already shaping the software market that every UK business operates in, regardless of sector or size.
When evaluating new software or renewing existing contracts, it is worth asking a different set of questions than the ones that have served well for the past twenty years. Not just what a platform does today, but how AI is integrated into its roadmap, what the pricing model looks like as AI usage scales, how autonomous actions are monitored and governed, and what the vendor's position is on accountability when AI-generated decisions go wrong.
Governance and oversight are particularly important. AI agents are genuinely capable, but they are not infallible. Businesses deploying autonomous systems need clear policies around what decisions AI can make independently, what requires human review, and how errors are caught and corrected. This is not a reason to avoid AI tools, but it is a reason to adopt them thoughtfully rather than reactively.
The businesses that will benefit most from this shift are those that develop a clear view of where AI can genuinely add value in their operations, understand the cost and governance implications before committing, and build the internal capability to manage AI-assisted workflows effectively.
This is exactly the kind of thinking we support through our AI services. Whether you are exploring AI tools for the first time, trying to make sense of the agent landscape, or looking to build a practical adoption plan that fits your business, having specialist guidance makes the difference between AI that genuinely delivers and AI that creates more complexity than it solves.