Finance One backs AI and automation to move faster

Non-bank lender Finance One is implementing its first AI strategy outline and framework, with the aim of finding operational efficiencies, improving service and moving faster with new products.

Finance One is part of the Investor Central Group, which also has a pair of debt collection businesses gained through acquisition.

Speaking to the iTnews Podcast, CIO Chris Doyle said that harmonisation, modernisation and automation are key themes of the company’s broader technology strategy and approach.

Where possible, technology effort has gone into establishing some common systems that can be used by the various entities within the group.

Legacy modernisation - largely complete -  has seen the technology team “unbundle code, undo direct integrations and make things more asynchronous so that we can have more robust processes, and then increase our performance by moving those things into the cloud,” Doyle said.

“[Thirdly], we're supporting the ambition of the business to build new products and move into new sectors, and the ubiquitous automation and AI discussion comes in [here].”

Doyle is taking a broad view of AI and automation.

The finance sector is no stranger to process automation and machine learning technologies, and Doyle sees these as continuing to have a role to play in AI discussions.

“When looking at automation [tools], there’s not one sledgehammer to crack a nut,” he said.

“It's about what different types of tools I have available to me, and there's plenty of vendors offering different levels of maturity and different types of solutions that can solve your problems.

“That's something that I've been directly involved in: working with the CEO and the executive to articulate what it is that we're actually dealing with? When looking to use [automation and AI] tools, how do you select them? How do you implement them? How do you make them successful?”

Doyle is particularly interested in adopting automation and AI tools that can help Finance One to build new products or to improve customer experience.

He “recently completed the AI strategy outline and the AI framework” for the group, and said implementation conversations internally “are going well.”

“They're positive. We've been running experiments of all different types using LLMs, putting in a platform for managing LLM access and having the appropriate … visibility of that centrally so that we can enable all the staff to use those tools, starting to enable areas of the business that have identified uses for agents and supporting them in how to build agents that are effective,” he said.

At a high level, the chosen approach - and the technology-based options employed to build automations - will differ based on the “criticality” of the process being targeted for improvement, and also the volume of work it handles.

For low-criticality, low-volume processes, the company has made vibe coding tool Lovable available to business users to create their own automations.

Vibe coding, Doyle said, offered a way for business units “to start to influence processes that typically don't get to the top of the priority stack” for the attention of the technology team.

“If you've got a low risk process that's low volume, just go and build something. It's fine.

“If the manager or team that owns the process isn’t going to get something built by technology, because it’s quite expensive internally [for us] to build things for you, then use [vibe coding instead].”

More elevated processes, particularly those around core or regulated aspects of the business, require more involvement of the central technology team when automation and AI use cases are being contemplated.

Call quality

One area where Finance One has embraced AI is in its contact centres, with Icana.AI’s CallCoach set up as an internal platform to review calls for quality assurance, sentiment and specific indicators such as hardship.

“We have somewhere north of 100 contact centre agents working in the collection space,” Doyle said.

“We've got other customer service and other roles too, but the bulk of people who are on the phone are in those contact centres, and we want all of them to have access to that capability.

All told, the company has call volumes of around 2000 a day and 10,000 a week.

It takes all “conversations” - defined as calls that run for at least one minute and adds them to an S3 bucket, from where CallCoach picks them up, analyses them and returns results to dashboards and reports.

Doyle said the AI had proven particularly useful in flagging indicators of hardship, which trigger regulated responses by financial companies.

“Hardship is a regulated process where if somebody meets the threshold for raising that they might have challenges, they don't need to use the word hardship. It’s about how we interpret which words they're using to say whether or not they're in financial distress,” he said.

“These are sensitive phone calls, and the hardship process in particular is something that the regulators are very keen on because they want to make sure that predatory lenders or predatory organisations are not out there causing pain.”

Running the AI means 100 percent of “conversations” can be analysed for hardship and other indicators, as opposed to just a small sample of calls where human teams alone do the review and analysis work.

With CallCoach, team leaders at Finance One are now presented with only the “five or 10” calls each day that require some sort of action or follow-up.

“So instead of looking at a sample of stuff that may or may not be useful, they're much more willing to spend time to look at the calls [flagged by CallCoach] because they know they're going to be useful,” Doyle said.

The response varies; it could be getting customers financial assistance, or recognising a well-handled conversation and using that as a training opportunity for other staff.

Doyle said that the company had seen “really tangible double-digit percentage improvements in compliance and engagement” since starting to use CallCoach.

Quality had improved, which also flowed through to revenue and the company’s bottom-line, he added.