At HLB's Technology & Innovation Symposium in Barcelona, Chief Innovation Officer Abu Bakkar told UKTN that artificial intelligence in professional services has reached a turning point. Firms that spent the last two years using AI to cut costs and save time are now being asked a harder question: can the technology actually improve the quality of audit and advisory work, not just the speed of producing it.
That distinction matters more than it sounds. A chatbot that drafts a client memo in minutes is a cost story. A system that catches an inconsistency in a set of financial statements before a partner signs off is a quality story, and it requires a completely different set of controls.
Why the efficiency framing falls short
Bakkar's argument, as reported by UKTN, is that firms treating AI purely as a productivity tool miss the bigger opportunity and the bigger risk. Audit and advisory work carries liability. A faster wrong answer is still wrong, and a faster wrong answer delivered to a client at scale is a bigger problem than a slow one.
For firms in the HLB network and beyond, the shift Bakkar describes means building applications with explicit quality objectives: fewer missed exceptions in a sample review, better detection of unusual transactions, more consistent application of a firm's own methodology across engagement teams. None of that shows up on a time-saved dashboard. All of it shows up in a peer review or a regulatory inspection.
What this means for SME and corporate advisory firms
Most mid-market advisory firms do not have HLB's symposium budget or a Chief Innovation Officer role. But the governance question Bakkar raises applies at any size: before a model touches a client deliverable, who reviews its output, what gets logged, and how does the firm know the error rate is not creeping up as usage grows.
Three practical steps for firms starting this work:
- Name an owner.AI oversight needs to sit with a specific partner or manager, not a committee that meets quarterly.
- Define the review gate.Decide, in writing, which AI-assisted outputs require a second set of human eyes before a client sees them, and which do not.
- Track errors, not just hours.A log of corrected AI output over six months tells a firm more about readiness than any usage statistic.
Bakkar's comments in Barcelona point to where the profession is heading: AI programmes judged on audit quality and client outcomes, not on hours removed from a timesheet. Firms that build the governance layer now will be the ones able to say, with evidence, that their AI-assisted work meets the same bar as their traditional work.
