AI governance // EP18

Ep18: New Level, New Devil

AI in professional services firms (Part 1)

Plenty of professional services firms have answered the "which tool?" question. A platform is chosen, a policy is drafted, the team has started training. Then comes the next problem: five people building the same workflow, agents nobody owns, and partners who can't say with confidence how AI touched the work going out the door. In part one of a two-part series, Rod Fernandez turns the interview chair around and asks Kim what she's seeing inside accounting, legal and HR firms heading into the last quarter of 2026.

Based on AI Your Business Podcast2026-09-23

The short version

The workshop that gets a team excited about AI is where a whole new set of problems starts. That's the "new devil" in the title: not the risk of doing nothing, but the risk that arrives once a busy team starts building on its own. Five people solve the same problem five different ways. Agents get built and nobody owns them. And when a client asks how AI touched their file, the partner can't answer with confidence.

Choosing a tool and drafting a policy isn't the finish line. It's where the harder governance questions begin.
A human in the loop reviews the output. A human in the lead owns it. Only one of those lets a firm stand behind the work.
Refusing to touch AI at all isn't safety either. Caution and safety are not the same thing.

Where firms stall

There's a specific stage of AI maturity where most professional services firms get stuck, and it isn't a technology problem. It's an unanswered governance question sitting underneath the work, and leaving it unanswered can set a firm back six to twelve months.

  • Several people building the same workflow, independently, with different assumptions
  • Agents created by whoever needed one, with no named owner afterwards
  • No clear record of how AI touched a piece of client work
  • A policy that exists, but wasn't built on decisions anyone actually made

What clients are about to start asking

Australia's Privacy Act disclosure changes take effect on 10 December 2026. The episode covers the client question every firm needs a confident answer to before then, and the places automated decisions hide that most firms wouldn't think to check.

  • Know where automated decisions are made in your own delivery, not just in the obvious tools
  • Give the decision an owner, so somebody can stand behind the output
  • Be able to describe, in plain words, how AI touched a client's work
  • Treat the answer as something you rehearse before you're asked, not after

How AI Collab thinks about it

The line worth keeping from this one is that caution isn't the same as safety. One actuarial firm in the episode has refused to touch AI at all, and that's still a position with risk attached to it. The way through is governance that's decided rather than assumed, which is exactly what the AI for Teams Toolkit and our governance work are built to produce.

This is part one of two. Part two continues the conversation.

Frequently asked questions

What does "new level, new devil" mean for AI adoption?
It describes what happens after the easy wins. Once a firm has picked a tool, drafted a policy and trained the team, a new set of problems starts: several people building the same workflow, agents nobody owns, and no clear answer to how AI touched the work that went out the door.
What is the difference between a human in the loop and a human in the lead?
A human in the loop reviews what the AI produced. A human in the lead owns the decision and the output. Only the second lets a firm stand behind the work, which is what professional services clients are actually asking about.
What do professional services firms need to answer before 10 December 2026?
Australia's Privacy Act disclosure changes take effect on 10 December 2026. Firms need a confident answer on where automated decisions are made in their work, including the places those decisions hide that most firms would not think to check.

Want to turn this into a practical AI plan?

AI Collab helps established Australian businesses move from interest to implementation: strategy, team capability, governance and custom builds that fit the way the business actually works.

Talk to AI CollabListen to the episode