The speed of AI development is producing markedly different responses from PKF-CAP and KLP. The former is introducing the technology within formal controls, while the latter has slowed its development work to assess how rapidly evolving audit applications could affect its plans.
Both firms are ACCA Approved Employers and recognise AI’s potential, but their approaches reflect different judgements about when and how to proceed.
Starting point
As a professional services firm, PKF-CAP handles substantial volumes of structured and unstructured information, including client documents, financial data, regulatory updates, workpapers, correspondence, and internal knowledge resources.
The firm identified high-volume, repetitive, and time-sensitive work as the areas most suited to the use of AI. These include audit, tax, accounting and outsourcing, client onboarding and customer due diligence, and internal operations.
“We see AI not as a replacement for professional judgement, but as a tool that helps our people manage complexity, respond faster, and focus more on higher-value analysis, review, and client engagement,” said Siew Inn Pong, Head of Business Solutions and International Liaison Partner at PKF-CAP. “In that sense, AI adoption is both an operational and strategic priority for PKF-CAP, supporting our competitiveness, resilience, and preparedness.”
According to Siew, the decision to adopt AI was not driven by trends, but by a recognition that the technology is reshaping how professional work is performed.
“We saw increasing pressure from rising client expectations, faster turnaround requirements, stronger competition, and the need to equip our people with capabilities that remain relevant in an environment where AI is widely used,” she said. “Maintaining traditional ways of working may still be possible in the short term, but standing still creates operational and strategic risk.”
KLP, meanwhile, is taking a more cautious approach. Lim Yeong Seng, the firm’s Managing Partner, said its use of AI is currently limited to research, drafting emails, and achieving small efficiency gains.
“We have only introduced individual tools, such as chatbots, to assist staff in their work. We’re still exploring how to implement a consistent, firm-wide AI strategy,” he said.
Transformation strategy
PKF-CAP has designed its technology stack around practical adoption, security, scalability, and integration with its teams’ existing ways of working. The firm uses enterprise productivity platforms such as Microsoft 365, including AI features such as Microsoft Copilot, alongside automation tools and selected third-party applications that incorporate AI into specific workflows.
“Rather than taking a one-size-fits-all approach, our current strategy is to combine established enterprise platforms with applications selected for particular tasks, including document processing, workflow automation, AI-assisted analysis, and productivity,” Siew said.
This, noted Siew, allows the firm to adopt AI in a practical and controlled manner while retaining flexibility as the technology and use cases continue to evolve.
She added that supporting the firm’s AI workloads involves more than technology. It also requires data security, access controls, governance policies, user training, and human review mechanisms.
“Client confidentiality and trust are critical, so our technology choices must be supported by responsible implementation and appropriate oversight,” Siew said. “The technology stack includes both a collection of tools and a governance framework, intended to improve productivity, responsiveness, and service quality while ensuring that AI is used securely, responsibly, and in accordance with professional standards.”
KLP, on the other hand, is taking a wait-and-see approach to AI. It previously moved away from robotic process automation (RPA) and towards software as a service (SaaS).
“RPA products are difficult to market because they can be easily copied,” Lim remarked. “SaaS is the way to bring products to market, and we’re developing a SaaS product, but AI for audit is advancing very quickly. We’ve therefore had to slow development while we monitor changes in the market.”
AI governance
As organisations adopt new AI applications, they must also establish governance frameworks to manage the associated risks, support responsible use, and maintain regulatory compliance.
In shaping its governance approach, PKF-CAP considered risks involving the confidentiality of client information, data protection, cybersecurity, the accuracy and reliability of AI-generated output, inappropriate reliance on AI, and compliance with professional and ethical standards.
The firm also considered its teams’ practical needs, including when AI may be used, what information must not be entered into AI tools, when human review is required, and when matters should be escalated.
“For PKF-CAP, the objective is to allow responsible adoption within clear boundaries. Governance isn’t intended to slow innovation, but to ensure that AI improves productivity, responsiveness, and service quality without compromising trust, professional standards, or client confidentiality,” Siew said. “Governance must advance alongside adoption. In a professional services environment, speed and innovation must be balanced with confidentiality, quality control, professional judgement, and human oversight.”
Implementation lessons
According to Siew, enterprises must remember that adopting AI is not merely a technology exercise. It requires changes in mindset, workflows, governance, and the way people work. Successful adoption therefore depends on whether teams understand AI’s purpose, recognise its limitations, and can use it responsibly in their day-to-day work.
“We also learned the importance of starting with practical use cases rather than trying to transform everything at once,” she said. “AI has been effective in addressing operational bottlenecks such as repetitive manual handling, document processing, drafting, summarisation, and initial analysis. Building from these practical use cases allows the firm to gain experience, improve adoption, and identify opportunities for greater integration over time.”
Siew also warned that wider use of AI could dilute technical knowledge. Incorrect or incomplete AI-generated output may be mistaken for an accepted technical position or standard approach.
“If staff rely on such output without sufficient review, it could lead to inconsistent application of professional standards and create risks in service delivery,” she explained. “To manage this, firms need to reinforce training, quality reviews, escalation channels, and a culture of critical thinking so that AI can support technical work without weakening the accuracy and integrity of professional knowledge.”
Disruption areas
Looking ahead, KLP recognises that AI is evolving rapidly and is therefore carefully assessing where the technology could make the greatest difference.
“We’re still at the exploratory stage with agentic AI, and we hope to introduce some of these applications across the firm,” Lim said.
For PKF-CAP, the next major area of disruption will be talent development and the preservation of professional knowledge.
“As AI becomes more embedded in professional services, the skills required of our people will also change,” Siew said. “Technical knowledge and professional judgement will remain important, but professionals will increasingly need to know how to work with AI tools, ask the right questions, review output critically, and apply judgement when using the technology.”
This will also affect talent retention, he noted, because well-trained professionals who are comfortable using AI may become more productive, adaptable, and valuable in the market.
“Firms that invest in AI adoption, technical training, and responsible use are likely to be more attractive to such talent, while firms that move too slowly may find it harder to retain people who want to develop alongside the technology,” Siew said.
At the same time, firms may face challenges if employees are not properly trained, she warned. Without adequate training and supervision, AI could be used ineffectively, staff could rely too heavily on its output, work quality could become inconsistent, and technical knowledge could be diluted.
“Staff may mistakenly treat AI-generated answers as generally accepted approaches if they lack sufficient grounding to challenge or verify them. The disruption is therefore not only about technology replacing tasks, but also about how firms develop, retain, and support their people while maintaining the accuracy, integrity, and depth of their professional knowledge,” she concluded.
