You Don’t Have an AI Problem. You Have a Trust Problem.

The real barrier to AI adoption for most Australian small businesses isn’t the technology. It’s the trust.

That’s the central argument from a conversation between Workday APAC CTO Shan Moorthy and Rita Arrigo, who helped establish Australia’s Responsible AI Network, published by InnovationAus this week.

Their point: organisations stuck on AI aren’t stuck because they lack access to tools. They’re stuck because nobody has answered a more basic question — what does success actually look like?

“The real gap isn’t about technology,” Arrigo says. “It’s about shared understanding of what success looks like.”

Sound familiar? If you’ve tried three or four AI tools in the past year and none of them stuck, this is probably why. Not because the tools were bad. Because you didn’t define the win condition before you started.

From tools to agents

The discussion focuses on a shift arriving faster than most small businesses realise: the move from generative AI (which helps you write and summarise) to agentic AI (which actually does things).

Early AI tools were co-pilots. You wrote a prompt, it gave you a draft, you edited it. The human stayed in the loop at every step.

Agentic AI works differently. You give it a goal and it figures out the steps. Book the meeting. Update the inventory. Follow up with the client. It’s not suggesting actions. It’s taking them.

That changes the stakes.

Moorthy puts it plainly: “This is where the real potential is. But it also introduces new questions around trust, accountability and governance.”

For a large organisation, those questions involve legal teams and IT departments. For a small business owner, they come down to something more personal: do I trust this thing enough to let it act on my behalf?

The pilot purgatory problem

Both speakers describe a pattern that will ring true for many small business owners: lots of experimentation, very little that makes it into daily use.

You try the tool. It works okay. You use it a few times. Then it gets busy, you forget about it, and it quietly disappears from your workflow.

The cause, according to the conversation, is almost never the technology. It’s the absence of a clear outcome driving the experiment. If you’re testing an AI tool without knowing what problem it’s supposed to solve, or what solved actually looks like, you’ll land in the same place every time.

Treating AI like a new team member

Moorthy’s reframe is worth sitting with. Instead of thinking about AI as software you deploy, think of it as someone you hire.

“IT becomes more like HR,” he says, “responsible for onboarding the agent, ensuring compliance, then handing it to the business to iterate.”

Scale that down to a business of five people and it still holds. When you bring on a new staff member, you don’t hand them the keys on day one. You start them on low-risk tasks. You watch how they handle it. You build trust in stages.

That’s how AI agents should enter a small business too. Start with tasks where a mistake costs you five minutes, not five clients. Get comfortable with how the agent works. Then, as trust builds, hand it more.

A practical trust ladder

Based on the risk framework the speakers describe, matching oversight to the level of autonomy, here’s how that plays out for a typical small business:

Low stakes to start with: drafting responses, summarising meeting notes, sorting inboxes, first drafts of social posts.

Once you’re comfortable: customer enquiry responses you review before sending, appointment reminders, invoice follow-ups.

When trust is earned: customer-facing communications that go out automatically, booking and scheduling, stock reordering.

The goal isn’t to stay at the low end permanently. Build to it.

The Australian context

Arrigo’s work with Australia’s Responsible AI Network adds a local angle. Australia is further along on AI governance thinking than many comparable economies, but most of that work targets enterprise.

For small businesses, the practical message is simpler: start with clear outcomes, match human oversight to the level of risk, and build from wins rather than betting on transformation.

“Trust is easy to lose and hard to gain,” Moorthy says.

For a small business operating on thin margins and close client relationships, that’s not a corporate governance principle. It’s just the reality of running a business.

Source: Agentic AI shifts the productivity debate from tools to trust — InnovationAus.com, 1 May 2026. The Productivity Levers series is produced by InnovationAus in partnership with Workday.

📖 More deep dives: News Deep Dives

📊 Compare AI tools side by side | 💼 Free resources & AI prompt packs