SCCU answers account balance inquiries with Agentforce Voice.

Provide scalable service tailored to customer needs.

For SCCU, maintaining a deeply personalised, customer-focused banking experience is a core priority. This is why the credit union considers its customer service centre a branch rather than a contact centre. In fact, it operates as SCCU’s largest branch, with some customers calling multiple times a day or week to check on recurring payments, such as pensions.

As the business continued to rapidly grow, the high reliance on phone support for routine, repetitive inquiries put pressure on the customer service centre. Wait times had escalated from a historical average of under two minutes to nearly eight minutes. The volume of calls limited the team’s ability to respond quickly to more complex and urgent customer needs, prompting them to explore solutions such as deprioritising certain calls. 

Agentforce Voice provided a more customer-focused alternative. By offloading simple account balance queries to an agent, the service centre could more quickly address more urgent or stressful calls such as reports of lost or stolen cards. However, SCCU needed a secure solution that would allow customers to converse naturally, as they did with the credit union’s employees.

Agentforce Voice delivers trusted support on customers’ channel of choice.

To meet the needs of customers who prefer phone-based banking, SCCU built and deployed a conversational agent named Nova using Agentforce Voice. The agent is integrated with the service centre's Vonage contact centre solution, which also features a new IVR menu. Now, customers can phone in, choose the option to obtain their account balance, and be immediately directed to Nova, who verifies their details before answering any questions.

Nova interacts with customers using natural language and with an understanding of different accents and phonetic variations. The agent is also trained to provide a human, empathetic experience and continuously analyses sentiment to recognise when a caller may be frustrated or distressed. It then immediately routes them to a live team member with full context.

Nova’s design includes sophisticated semantic intelligence, which helps ground this conversational interface in real-world banking needs. For example, because many credit union customers don’t memorise their account numbers, they refer to them using product codes or nicknames. Nova is trained to understand these and retrieve the correct account balances across different accounts.

Customers have quickly accepted Nova as part of the service centre team, with many speaking with the agent multiple times a day to check their balance for expected payments. “We’ve 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,” said Jessica Jackson, Chief Customer Officer, Southern Cross Credit Union.

SCCU has set a target for Nova to successfully identify and verify 80% of callers and successfully provide account balances to 95% of those verified. While it is still working towards the verification target, it has reached its target of providing account balances to 95% of those verified.

In addition, Nova successfully handled 7% of customer calls in one of the first few months post-deployment. This is significant given that the agent currently only handles balance inquiries. Also, while these interactions are typically shorter and less complex than other call types, the reduction in workforce effort equated to about 4% of total staff hours. Overall, this demonstrates Nova's early value and highlights the potential for greater efficiencies as additional use cases are introduced.

To continuously measure and improve Nova's performance, SCCU relies on dashboards in CRM Analytics to track key metrics like the number of calls connecting to Nova, the percentage of callers verified, and the percentage of callers obtaining their account balance. They are also tracking escalations both pre- and post-account balances being obtained.  

The service centre team additionally reviews call transcripts to understand customer sentiment and why some callers may escalate to live agents. These insights will enable the team to make iterative improvements to Nova, and ultimately increase the number of calls resolved by the agent.

SCCU is now planning to deploy a 'click to chat' agent on its website to respond to general inquiries such as questions about branch hours. The agent would use Data 360 to retrieve answers from knowledge articles, ensuring accurate answers grounded in SCCU’s data. Leveraging the same data and learnings from that project, the credit union also plans to expand Nova's voice capabilities to handle similar questions.

Service Cloud provides a single source of truth across the organisation.

The launch of Agentforce extends SCCU’s use of Salesforce, which it first rolled out to its service centre four years ago with Salesforce Voice. Today, 99% of the organisation utilises the platform, with Service Cloud acting as a single source of truth for customer data. Teams across nearly every department use the platform daily to track and manage activities such as cases, complaints, deceased estates, and fraud and scam reports.

To keep this single source of truth up to date, SCCU uses MuleSoft to manage APIs that link Service Cloud directly to its core banking system and other key platforms such as its online loan application system.

Replacing legacy five-minute batch database syncs, MuleSoft connectivity allows customer records to be created and updated instantly, and for online loan application updates to be ingested in real time. MuleSoft is also used to ingest industry home loan rates into Salesforce, automatically refreshing daily dashboards in CRM Analytics that inform decision-making.

SCCU is one of the first to go live with Agentforce Voice in Australia and New Zealand, and to design a human and trusted experience, the credit union engaged Salesforce Forward Deployed Engineers (FDEs). Working alongside the FDEs, they built, tested, and deployed Nova in just six weeks. This included building a robust process to verify account holders that is much more secure than traditional methods, such as one-time passcodes. Additionally, they’ve built in a safeguard that routes some customers to a human rather than Nova based on certain criteria or vulnerabilities.

To deliver these capabilities securely in a regulated banking environment, the FDE team structured Nova's architecture to prevent LLMs from directly accessing customer data. Instead, Nova acts as a secure front-end interface, collecting customer inputs one at a time to reduce speech-to-text translation errors and passing them to backend tools connected to Service Cloud. The agent similarly uses deterministic variables to fetch and read out account balances.

For SCCU, the key differentiator of Salesforce has been the flexibility of Agentforce to build a custom, high-trust agentic experience that meets the unique needs of the credit union and its customers. Rather than forcing the business into rigid, pre-packaged software parameters, Agentforce gave the SCCU team the flexibility to co-design and configure Nova's personality and instructions.

"Sometimes a third-party provider will just say, 'This is our solution, and you need to work within its parameters,” said Melissa Marsh, Salesforce Platform Manager. "Our experience working with Salesforce is that they were happy to work with us and build something that fit into our business.”

SCCU unlocked Agentforce's flexibility through close collaboration with Salesforce Forward Deployed Engineers (FDEs). Partnering with the FDE team, SCCU leveraged Agent Builder to construct and iterate on Nova's capabilities. Agent Builder's low-code interface let the team easily map specific topics, define automated actions, and write precise instructions. The team also leveraged determinism within Agent Builder to control actions and outputs, rather than leaving conversational paths open to Large Language Model (LLM) interpretation.

By combining these deterministic guardrails with the built-in protections of the Agentforce Trust Layer, SCCU also created an agentic experience that adhered to strict financial regulatory standards.