AI: a larger firm’s perspective

Conversations around artificial intelligence (AI) in the audit profession have evolved rapidly from theoretical possibility to practical reality. As highlighted during the faculty webinar AI in audit: opportunities and challenges, firms of all sizes are now actively deploying AI tools across their audit processes. Despite the excitement, there remains considerable confusion around what the term AI encompasses and what responsible AI adoption means for audit.

Putting AI into perspective

When we talk about AI in audit, we include a broad spectrum of technologies that many firms are already using, often without realising it. Software that uses AI techniques has long been quietly revolutionising our work and includes:

• optical character recognition (OCR) in tools that automatically extract, cross-reference and validate financial documents – accelerating audit procedures, such as substantive tests of detail;
• pattern recognition in data analytics platforms – deepening the auditor’s understanding and analysis in complex high-risk areas such as revenue recognition; and
• automated anomaly detection in various audit software types – enhancing the auditor’s risk assessment and responses such as journal entry testing.

What has captured attention recently is generative AI (Gen AI), which includes tools such as ChatGPT and Microsoft Copilot. These systems work by predicting ‘what comes next’ with high probability, making them powerful assistants for content creation, analysis, and problem-solving. However, it’s crucial to understand that they produce unique results each time and can generate convincing but incorrect information (known as hallucinations).

This technology offers unprecedented opportunity for all firms, but with this comes significant responsibility. Larger firms, in particular, must grapple with complex client relationships, regulatory obligations across multiple jurisdictions, and quality standards that demand careful consideration of how and where we deploy these tools in our audit practices.

Advanced applications for larger firms

At RSM and other large firms, the approach to digital audit capabilities may centre on four key application areas.

1. Advanced pattern recognition and analytics

Instead of sampling, firms analyse entire populations of transactions. This allows them to identify unusual patterns across millions of records that would be impossible to detect manually, while other tools analyse revenue transaction flows through the sales cycle, surfacing outlier transactions for further analysis.

2. Process automation and integration

Large firm solutions enable data extraction and integration, connecting with supported accounting systems. This automated approach means client teams can focus on other deliverables. Firms may also implement AI-driven document extraction capabilities that review financial statements and supporting documentation, while keeping the human firmly in the loop.

We’re also seeing intelligent workflow orchestration across audit teams, with automated tracking of engagement progress and real-time updates on audit status. This isn’t just about efficiency – it’s about consistency and quality as well as client and audit team experience.

3. Sophisticated analytical solutions

Custom solutions for specific audit challenges include tools that leverage advanced algorithms for complex calculations such as lease accounting and inventory valuation. These solutions integrate with existing audit technology platforms, enhancing rather than replacing current tools and maintaining consistent methodology.

4. Emerging agentic capabilities

We’re beginning to see the emergence of agentic AI systems – AI that can perform sequences of tasks autonomously. While still in development, these systems show promise for navigating complex processes with minimal human intervention, making decisions about what information is needed and following logical paths to gather required evidence such as within client deliverable environments. This technology could eventually free up auditors to focus on complex judgements and value-added insights rather than mechanical data gathering.

Navigating challenges and human-centric implementation

Implementing AI isn’t without challenges. At RSM, we’ve adopted what we call a ‘human-centric AI integration’ approach, emphasising the ‘human in the loop’ philosophy. While AI tools enhance our capabilities, all work is reviewed and signed off by qualified audit team members.

Technical limitations remain. Effective prompting techniques, with carefully crafted instructions or questions for AI, can reduce risks, but critical verification remains essential for audit use cases. During the recent faculty ICAEW webinar on AI in audit, I shared the AUTOMAT structured framework for effective prompting, which can significantly improve AI output quality and consistency.

Data protection and security present perhaps the greatest challenge for firms of all sizes. Client confidentiality obligations remain paramount, and careful vendor selection based on data usage policies is crucial. At RSM, we’ve established clear guidelines: 

• Only Gen AI tools with enterprise data protection (EDP) may be used where prompt data is confidential.
• Personal data should not be entered into AI tools without a Data Protection Impact Assessment having been completed first.

Governance and digital mindsets

Professional judgement exercised with scepticism and ethical consideration remains fundamental to auditing and central to auditors’ knowledge, skills, and processes. Successful AI and technology adoption requires comprehensive governance frameworks underpinned by the following digital mindsets: 

• Data-driven (putting data at the centre of our work)
• Interactive (delivering interactive and engaging outputs)
• Knowledge-centric (sharing knowledge across teams)
• Human-focused (prioritising client relationships) 
• Future-smart (continually developing digital skills)

Many firms have core principles that guide AI usage. These range, for example, from actively encouraging AI use while maintaining accountability, to ensuring all outputs are reviewed with critical judgement. This may involve establishing centres of excellence: specialised teams that develop best practices, support complex implementations, and provide training across the firm.

Investing in continuous learning has been key to our success. Regular training sessions and university partnerships for advanced analytics skills, enable us to blend human expertise with cutting-edge technology effectively.

Looking to the future

The regulatory landscape is evolving rapidly across jurisdictions. Standard-setters and regulators worldwide are considering, or have issued, guidance on AI use in audit, such as the AI in audit – Illustrative example and documentation guidance published by the UK’s Financial Reporting Council in June 2025.

As AI, the profession’s use of it and regulatory guidance on this continue to evolve, there will be opportunities for all firms to differentiate through technology adoption while maintaining professional standards. 

Conclusion

AI augments rather than replaces professional judgement. The future of audit is about technology and professional judgement working in concert to deliver higher quality audits and better client service. 

Core audit principles – professional scepticism, evidence-based conclusions, understanding client business and risks – remain unchanged. What changes is our ability to apply these principles more comprehensively and effectively than ever before.

Successful AI implementation requires more than just deploying tools. It requires a culture of continuous learning, clear governance frameworks, and commitment to keeping humans at the centre of the process. 

The AI revolution in audit is not coming – it’s here. The question for firms is not whether to embrace AI, but how to do so responsibly, effectively, and in service of our ultimate goal: delivering high-quality audits that serve the public interest.

Konrad Bukowski-Kruszyna, Head of Audit Data Analytics, RSM UK