AI in a variety of forms has continued to make its presence felt in accountants’ lives, enabling them to approach their work in different ways.

Development and use of AI agents has significantly increased this year, driven not just by accounting firms, but software vendors who are incorporating agents into audit, accounting and finance software.

Big Four, big ideas

With their considerable internal resources, the Big Four firms have unsurprisingly surged ahead in accountancy’s AI league.

In March, PwC launched an entire agent operating system, agent OS. Capable of integrating with a range of leading enterprise solutions such as Oracle and Salesforce, the OS connects custom and pre-built AI agents into business workflows across a variety of industries.

Three months later, KPMG unveiled Workbench, empowering its workforce with a network of AI assistants and chatbots. In July, Deloitte announced a set of new AI capabilities within its cloud-based audit and assurance platform Omnia, with Gen-AI tools providing facilities for enhanced document review and improved navigation of financial statements.

As the year drew to a close, EY announced its EY.ai Tax Labs and EY.ai Tax Agent Factory tools, designed to link tax advisory with other business functions – such as compliance, operations and value creation – to deliver greater strategic value from tax.

Mid-tier manoeuvres

However, the Big Four were not the only sources of Gen-AI innovation within the accountancy sector. As highlighted in ICAEW’s AI masterclass videos throughout 2025, many mid-tier firms are identifying and implementing use cases for Gen-AI. In the summer of last year, mid-tier firm Moore Kingston Smith (MKS) outlined what it has achieved with its Gen-AI platform AssureRight, which makes employees’ day-to-day workflow more efficient, holistic and easier to manage. As MKS Director of Innovation Jared Goodrich pointed out, the platform is the result of a long-term focus on technology development at the firm that has enabled it to develop a Gen-AI system internally.

In November, MKS’s efforts garnered further coverage at Reuters. According to the firm’s Head of Digital Transformation Becky Shields, using Gen-AI to build bespoke solutions allows organisations to “do a lot with a little”. She also described the cost as “pennies in the pound,” compared to spending on other types of technology.

Muted returns?

That said, questions are hanging over the health of return on investment (ROI) from AI expenditure. In a recent survey of almost 2,000 senior executives in Europe and the Middle East, Deloitte found that most respondents are achieving satisfactory ROI on a typical AI use case within two to four years – significantly longer than the typical payback period of seven to 12 months expected from other types of technology investments.

Meanwhile, a report from MIT said that despite $30bn to $40bn of global enterprise investment in Gen-AI, 95% of organisations are seeing zero returns. Indeed, it noted, ROI from Gen-AI innovations is often highest from back-office departments such as operations and finance – but budgets are frequently biased towards more visible, front-office functions.

Glitches persist

Amid the constant stream of hype, it often seems hard to confront Gen-AI’s flaws – but mistakes do happen. In October, The Guardian reported that Deloitte had agreed to pay a partial refund over errors in a report that the firm had written for the Australian Government. Deloitte admitted that it had used Gen-AI during the production of the AU$440,000 report – and bogus, hallucinated references had crept into the final draft.

The following month, Deloitte was once again in the spotlight, with similar concerns swirling around a report it had prepared for the Department of Health and Community Services in Newfoundland. That one was budgeted at $1.6m.

Junior freeze

Gen-AI is also reportedly having a chilling effect on the prospects of career entrants. In December, the Financial Times warned that starting salaries for consultants in the Big Four have stagnated since 2022. Indeed, a senior Big Four Executive told the paper: “Employment costs are going up, because of national insurance, the minimum wage etc, and you might be in a better place investing in AI and offshoring than in people.”

Adoption hurdles remain

Several barriers to AI adoption have been identified, including skills shortages, concerns about AI risks and the regulatory environment. These will need to be addressed for wider-scale adoption in 2026.

Hero of the hour?

One factor that could make a difference to accountancy’s relationship with AI next year is that we now have someone tasked with nurturing that relationship.

In November, Chancellor Rachel Reeves appointed Shaheen Sayed, Chief Commercial Officer for Reinvention Services at Accenture, as AI Champion for the professional and business services (PBS) sector.

Sayed will work closely with businesses and government to accelerate AI adoption, and will help PBS firms harness technology to boost productivity and growth. Sayed is eager to get cracking – in a LinkedIn post to mark her appointment, she hailed a “pivotal opportunity” to help position the UK as “the world’s most trusted adviser to global industry by 2035”.

Getting to grips with risk

Another likely trend for 2026 that has its roots in 2025 is a tighter focus on risk management.

In June, the Financial Reporting Council (FRC) published landmark guidance on the use of AI in the audit profession. While the guidance was not prescriptive, it provided a clear signal that the watchdog was keeping a close eye on the potential risks of AI in audit. FRC Executive Director of Regulatory Standards Mark Babington acknowledged: “AI tools are now moving beyond experimentation to becoming a reality in certain audit scenarios.”

ICAEW Head of Tech Policy Esther Mallowah says that outside of skills, AI adoption in the profession is hindered by risk and how to manage it. “Part of that riddle will be addressed through regulation – but not all of it,” she says.

“In 2026, I expect organisations will focus more on how they govern the use of AI – both internally, and in conversation with their vendors – and that links to assurance, too. Governance, regulation and assurance are all interconnected, and will be really important for driving adoption next year.”