5 lessons from a fintech’s AI

Ask any bank, insurer, or payments company where AI pilots stall, and the answer is rarely the model. It’s onboarding — the sprawling, compliance-heavy process of verifying who a customer is, which products they qualify for, and which internal systems need to know about it. Financial services has more regulatory checkpoints per customer journey than almost any other industry, which makes it both the sector where AI pilots stall most often and the one where getting orchestration right pays off fastest.

Here’s what one payments company’s fix looked like, and the 5 lessons other finance firms can learn from it.

1. The problem is rarely the AI — it’s what’s underneath it

Novatti, an ASX-listed payments company operating across Australia and New Zealand, ran into a version of this problem a lot of FSI CIOs will recognize. The company had grown through four separate business units, each with its own onboarding process, systems, and compliance checks. This was a structure that made sense while scaling, but became a serious drag the moment leadership wanted one consistent customer experience across all four.

2. Design for scale from day one, not after it hurts

CEO Mark Healy’s mandate for the rebuild was blunt: the process had to hold up whether the company had its current customer volume or a hundred times more. “It should be exactly the same effort,” is how he put it, regardless of scale.

3. Connect the systems before you add intelligence on top

The fix wasn’t a new AI model bolted onto the front end. It was orchestration underneath: connecting the CRM, verification systems, and legal document workflows so information entered once flowed automatically to every system that needed it, instead of being re-keyed at each handoff. Onboarding now verifies a customer’s organization at first contact, before collecting product-specific details, and additional information gets added later without repeating verification already completed.

4. Use onboarding as your AI-readiness test case

Onboarding friction is a documented problem across financial services broadly, and every FSI institution running multiple product lines, legacy systems, and a compliance function that needs a clean audit trail is running some version of the same architecture problem. That’s also why onboarding is the right proving ground for AI orchestration and broader digital transformation efforts, ahead of flashier use cases like agent-driven service or fraud detection. Get the identity, access control, and system-of-record connectivity right here, and every AI use case that follows — document processing, compliance monitoring, personalized recommendations — inherits a governed, connected environment instead of starting from scratch.

5. The real win was time, not just speed

Engineers who used to spend their days maintaining point-to-point integrations between onboarding systems got that time back for the work that actually moves the business forward. The CRM became a genuine single source of truth, which meant sales stopped chasing down document status manually, signed agreements triggered instant notifications instead of someone checking inboxes, and legal documents lived in one searchable place instead of scattered across four separate archives. Every one of those is a small amount of time returned to teams who used to lose it to busywork — and across a company, that adds up faster than the headline speed number suggests.

If you’re looking for the highest-leverage place to prove out AI-ready orchestration, don’t start with the flashiest use case. Start with the process that already touches the most systems and carries the most compliance weight. Solve the plumbing there, and everything downstream moves faster.

Want to see how Novatti’s onboarding runs today? Read the full story or talk to our team about your own onboarding architecture.