AI agents are coming to back-office operations — new research shows the future

The data is in, and it confirms what forward-thinking business owners have suspected: AI agents are no longer confined to software development teams and tech companies. According to new research from Anthropic, one of the world's leading AI companies, back-office automation now accounts for 9.1% of all agentic tool calls — making it the second-largest category of AI agent deployment after software engineering.

For UK SME owners who have been watching the AI revolution from the sidelines, wondering when it might reach their operations, the answer is clear: it already has. The question is not whether AI agents will transform back-office work, but how quickly businesses will adopt them to stay competitive.

What the research actually shows

Anthropic's comprehensive study, titled “Measuring AI agent autonomy in practice,” analysed millions of human-agent interactions to understand how AI agents are actually being deployed in real-world scenarios. The findings reveal a landscape far more diverse than the common perception of AI as a tool primarily for developers and data scientists.

While software engineering still dominates at 49.7% of all agentic activity, the remaining half is spread across traditional business functions. Back-office automation sits at 9.1%, followed by sales and CRM (4.3%), finance and accounting (4.0%), data analysis and business intelligence (3.5%), customer service (2.2%), and e-commerce operations (1.3%).

Combined, these non-software applications account for approximately 50% of all AI agent usage — and the trend is growing. What makes this particularly significant for SMEs is that these are precisely the areas where small firms typically struggle with manual processes and limited resources.

The human-in-the-loop reality

One of the most important findings from Anthropic's research addresses a common concern among business owners: the fear that AI agents operate as completely autonomous systems, making critical decisions without human oversight. The reality is far more reassuring.

The study found that 73% of all AI agent tool calls involve some form of human oversight, and only 0.8% of actions are truly irreversible. These are not rogue systems making independent decisions about your business. They are sophisticated assistants that handle routine work while keeping humans firmly in control of important outcomes.

As Anthropic's researchers put it: “Effective oversight doesn't require approving every action but being in a position to intervene when it matters.” This aligns perfectly with how successful SMEs actually want to work — maintaining control while eliminating the time-consuming manual tasks that prevent growth.

Why UK SMEs should pay attention now

The research reveals that we are in the early stages of a significant shift. Software engineering dominated early AI agent adoption because developers were the first to build and deploy these tools. But as the technology matures and becomes more accessible, adoption is spreading rapidly into traditional business operations.

For UK SMEs, this presents both an opportunity and a timing challenge. The firms that adopt AI agents now will have a significant advantage over those who wait. Consider the economics: if back-office automation can reclaim even a portion of the 40 hours per week that typical small firms lose to administrative tasks, the productivity gains translate directly into competitive advantage.

UK SMEs face particular pressures that make this technology especially relevant. Labour costs are rising, skilled staff are increasingly difficult to find, and regulatory compliance requirements continue to expand. AI agents can address all three challenges simultaneously — handling routine work more efficiently than manual processes, reducing dependence on hard-to-find specialist staff, and maintaining consistent compliance standards.

What this means in practice

The categories identified in Anthropic's research map directly onto the challenges that UK SMEs face daily. Back-office automation includes invoice processing, document handling, compliance monitoring, and workflow coordination. Sales and CRM encompasses lead qualification, follow-up sequences, and customer communication. Finance and accounting covers transaction processing, reconciliation, and reporting.

These are not theoretical applications. They are live, working systems handling real business processes for firms that have made the transition from manual to automated operations. The 9.1% figure for back-office automation represents thousands of businesses that have already discovered what AI agents can accomplish in practice.

What makes this particularly compelling for SMEs is that the barriers to entry continue to fall. The same research shows that most AI agent deployments include safeguards and human oversight, making them far less risky than early adopters might have assumed. Modern platforms designed specifically for small businesses can deliver enterprise-level automation without enterprise-level complexity or cost.

The timing advantage

Anthropic's research captures a moment in time when AI agent adoption is expanding beyond its initial domains but has not yet reached mainstream adoption in traditional business operations. For SMEs, this represents a window of opportunity.

The firms that implement AI agents now will have time to optimise their workflows, train their teams, and establish competitive advantages before the technology becomes ubiquitous. More importantly, the study demonstrates that AI agents complement rather than replace human expertise. The highest-performing deployments combine AI efficiency with human judgment, which is exactly how successful SMEs prefer to operate.

If your firm is struggling with the administrative burden that prevents growth, or if you are looking for ways to improve efficiency without adding headcount, the research from Anthropic provides compelling evidence that AI agents can deliver real results in back-office operations. The technology is ready, the applications are proven, and the timing may not get better.

The question for UK SME owners is not whether AI agents will transform back-office work — the data shows they already are. The question is whether your firm will be among the early adopters who gain a competitive advantage, or among those who catch up later.

Source: Anthropic Research: Measuring AI agent autonomy in practice, February 18, 2026