Southern Cross Credit Union Goes Live With Agentforce Voice, Among the First in ANZ
Southern Cross Credit Union (SCCU), a customer-owned financial services provider, is one of the first organisations in Australia and New Zealand to go live with Agentforce Voice. Its voice agent, Nova, now provides account balances to 95% of customers who select the balance enquiry option and are successfully verified. In just its first few months of operation, Nova is also handling 7% of all customer calls.
The Urgent Need for a Smarter Voice Experience
SCCU treats its service centre as its largest branch, not a contact centre. Some customers call several times a day or week to check on recurring payments such as pensions. As the credit union continues to grow, routine calls pushed average wait times from under two minutes to nearly eight. That made it harder for the team to get to urgent calls, such as lost cards or scam reports.
“We needed to allow customers in stressful situations, like reporting scams, to get straight to a human without waiting,” notes Jessica Jackson, Chief Customer Officer, Southern Cross Credit Union. “But we’ve also found a lot of our regular customers who frequently call for account balances love speaking with Nova, even some of those we assumed might be hesitant.”
How Nova Delivers Secure, Human-Centric AI Support
When a customer calls SCCU, they choose the balance option from a new menu and are connected to Nova, which is built on Agentforce Voice and integrated with SCCU’s Vonage contact centre solution.
Before sharing anything, Nova verifies the caller’s details, and SCCU says this is more secure than traditional methods such as one-time passcodes. Nova is able to understand different accents and phonetic variations, and recognises the product codes and nicknames customers use instead of account numbers, so callers can talk the way they normally would.
Nova also monitors sentiment throughout the call. If a customer sounds frustrated or distressed, she routes them to a live team member with the full context of the conversation, so they don’t have to repeat themselves, with separate safeguards in place to send some callers to a live team member based on set criteria or vulnerabilities.
In a regulated banking environment, how Nova handles data matters as much as how she sounds. Her architecture stops the LLM from directly accessing customer data. Nova collects inputs from the caller one at a time, which reduces speech-to-text errors, and she then passes them to backend tools connected to Service Cloud, using deterministic variables to read out balances accurately.
“We see Nova as a team member, and we’ve used Agentforce to give her a personality and configure her to be exactly like we want her to be, which is just like us,” says Melissa Marsh, Salesforce Platform Manager, Southern Cross Credit Union. “She sounds like us, and our customers treat her that way.”
From Concept to Launch in Six Weeks
The SCCU team worked with Salesforce Forward Deployed Engineers (FDEs) to build, test and launch Nova in just six weeks. Starting with workshops at SCCU’s head office to shape Nova’s tone, including how much humour and empathy she should show, Nova was then tested within an Agentforce Testing Centre, which helped simulate different accents and dialects, before subsequently being rolled out to employees first. The team used Agent Builder’s low-code interface and deterministic controls, together with the Agentforce Trust Layer, to meet financial regulatory standards.
“The FDEs felt like part of our team,” continues Jackson. “We met daily and had constant communication via Slack, which provided us with continuous visibility of the project. Sometimes a third-party provider will just say, ‘This is our solution, and you need to work within its parameters.’”
“Our experience working with Salesforce is that they were happy to work with us and build something that fit into our business.”
Following an impactful first few months, the SCCU team is now working toward its target of verifying 80% of callers, providing more room for Nova to take on increased volume as verification rates improve.This will also lead to a ‘Click to Chat’ website agent for general questions such as branch hours, drawing on Data 360 to pull answers from knowledge articles, with plans to extend Nova’s voice capabilities to similar questions.
The Salesforce View
“Southern Cross Credit Union shows that voice AI can work in a regulated, trust-critical industry when it is designed around people,” says Stephanie Wells, AVP, Salesforce. “By pairing deterministic guardrails with natural conversation, Nova handles routine calls so SCCU’s team can focus on customers who need them most.”
“This is not an overnight success; the implementation of Agentforce builds on four years of Service Cloud use at SCCU with 99% of the organisation using the platform, and with MuleSoft linking it to the core banking system in real time. SCCU’s speed to value with Agentforce Voice is a direct result of getting its data, integrations and people right first.”
“When the foundations are already in place, AI doesn’t need to be a long journey. It becomes the next step,” concludes Wells.
