Inside Colonial First State’s AI pilot with Avanade

Project focused on applying AI to customer service, advisor support and operational efficiency to improve outcomes.

Colonial First State (CFS) has enlisted Avanade to roll out a trio of AI pilot programs across customer experience, financial advisers and day-to-day workflows.

All employees have access to generative AI, with three-quarters of permanent staff using Microsoft Copilot for daily work. Colonial wanted to go further and following brainstorming sessions, developed three custom proof of concepts (POCs) to explore potential applications.

Daniel Arico, generative AI product owner at CFS said, “Our exploration of generative AI was really driven by an unprecedented surge of enthusiasm from our employees.”

The superannuation and investment industry is a highly regulated, complex and competitive space and having partnered over the years, Colonial turned to Avanade to explore generative AI use-cases.

Financial services requires specialised tools that can navigate complexity while maintaining human accountability to review the work and responses and decide on the appropriate action, noted Saurabh Verma, financial services lead at Avanade Australia.

Colonial AI pilots explore ways to improve customer and employee experiences

A customer feedback pilot trialled Azure OpenAI and traditional machine learning to improve insights from customer complaints by identifying trends and common issues, recategorising responses and sharing reports across the organisation using Microsoft Power BI.

“With Gen AI, we analysed unstructured data from the free-text fields and could recategorise complaints and surface a completely different level of insight that wasn’t possible before,” said Verma.

For the second project, Azure OpenAI with GPT-4 was used to develop a retrieval-augmented generation (RAG) chatbot, known as FirstTech. It was deployed to search and summarise technical services guides and select webpages to assist financial advisers.

This was far more specialised than an off-the-shelf Copilot tool. CFS needed something that could work across multiple data sources and provide specialised, domain-specific insights, Verma told CRN Australia.

The third project involved building a custom chatbot with Azure OpenAI and GPT-4 to search superannuation and investment databases and quickly locate information for customer guidance teams.

“This helps customer service teams accelerate the rate at which they can find information accurately, reducing call handling times and improving first-time call resolution,” he said.

Once in full production, the tool has the potential to improve productivity, enhance training for new agents, lower average call handling times and reduce calls to the technical support team by 40 percent, according to CFS.

Partners the essential ingredient in successful generative-AI adoption

The Colonial pilot program shows how businesses turn to partners to support innovation through new technologies like AI.

“We understand it’s not just a technology change. It’s an organisational and people change, and it’s building up the tool and process capability to help organisations through that,” Verma said.

From the board down to leadership, there’s now a strong belief that generative AI is a mainstay and the conversation has shifted from comparing different language models to applications and benefits.

“They’re not just looking at GenAI to take out costs or make humans more productive. Organisations are starting to look at how to use the technology to grow and improve revenue,” he said.

Verma believes that partners can support organisations to broaden their perspective about applying AI to areas that can make the greatest impact.

“Most organisations want to start with an internal process challenge and create a safe space to try the technology before opening it up to the customers. … Where we see really good feedback and results is when you start opening it up to customers,” he told CRN Australia.

Overall, partners play an important role helping organisations turn curiosity into use-cases, guiding decisions and providing expert input.

“We're finding organisations need to learn about the kind of capabilities, people and tooling they need,” he ended.