Making the most of advances in insurance system technology

Advancements in AI and data processing are already reshaping the way brokers, underwriting agencies, and insurers operate. Many have rushed to integrate AI and automation, inspired by the promise of greater efficiencies and access to real-time data.

The challenge, however, lies in harnessing these developments while maintaining the stability and integrity of core systems—something BAIS insurance technology has been navigating for decades.

BAIS has long taken an innovative approach to insurance software, consistently working ahead of industry trends. It introduced cloud-based solutions before “cloud computing” became a mainstream term and developed search tools that were later replicated by competitors.

The company’s iBAIS 2.0 platform, used by brokers and underwriting agencies across Australia and beyond, has been another such innovation, streamlining policy management, underwriting, claims, and compliance. Now, as it prepares to launch iBAIS 3.0, the focus is on enabling insurance professionals to make better use of their data through AI-driven efficiencies—without compromising security.

CEO David Hampton says a key challenge with new technology is ensuring that data remains accurate and structured correctly. The integrity of the core operating system is the livelihood of insurance brokers and underwriting agents—the “golden record,” as it’s known at BAIS.

“If the bank balance doesn’t balance, you can’t do a bank reconciliation. You’re going to have trouble,” Hampton says. “Some systems have a one-sided journal entry. Accounting is double-entry—debits and credits have to go somewhere.”

Data errors, Hampton explains, often originate not from the technology itself but from poor data entry or inconsistent coding. “The problem is that classic, ‘garbage in, garbage out,’” he says.

As the industry explores third-party AI tools, the risk of data corruption increases.

Hampton warns that some brokers are turning to AI tools such as ChatGPT without fully considering the consequences. “People are sending client data out to ChatGPT. But that information is very sensitive,” he says. “Once it’s in the cloud, it’s in the cloud forever.”

Smaller brokers, he notes, may be particularly vulnerable, as they lack the IT infrastructure of larger firms to mitigate risks.

Outside of the external cloud-based systems, BAIS sees AI as a tool that can enhance human decision-making rather than replace it. The company’s virtual assistant model within the upcoming iBAIS 3.0 is designed to help brokers and underwriters retrieve data faster without needing complex queries or reports.

“We don’t think artificial intelligence is ready to start making decisions,” Hampton says. Instead, iBAIS 3.0’s virtual assistant gives people access to their own data in a way that they’ve never had before. “We’ve developed tools where, just like chat GPT, you can talk to it on any device, from anywhere.”

Sheri Andrews, Head of Client Services and Marketing, says this is particularly valuable for executives who may not use these systems daily, without having to remember how to navigate the system and generate reports.

“It’s about bringing 100% of their data to life,” she says. The majority of professionals outside of IT and data analytics use their business tools to a minute percentage of their capabilities. Of the 100s of things that these tools can do, most people do 10.

BAIS 3.0’s AI-driven system helps users find valuable data by asking simple questions in plain language, such as how many policies exist in a post code or cresta zone, and receiving answers instantly.

“If you’re able to query your own data at any point in time, you can just say, ‘give me a list of all policies lapsed in the last three years that I should be talking to again,’ and you get that information, short and sharp,” explains Andrews.

A significant focus of iBAIS 3.0’s development has been ensuring AI-generated responses remain reliable. Hampton acknowledges that many mainstream AI models today produce inaccurate or misleading answers—a phenomenon known as “hallucination.”

To minimise this inaccuracy, BAIS teaches the machine to look for answers only within that database. If it doesn’t locate an answer, it says so.

BAIS has also built safeguards into its AI system to protect sensitive client data. Andrews explains that before any data leaves the system, it is anonymised.

“If it finds a name, for example, it will modify that to a fictitious name, and if it finds an address, it’ll modify it to a fictitious one so that nothing sensitive leaves our boundaries,” she says.

As BAIS prepares to roll out iBAIS 3.0 from May, the company continues to think about what comes next. Hampton sees a future where insurance platforms actively support decision-making, not just store records.

“The goal is to move from passive systems to interactive ones. Imagine sitting down in the morning and having the system tell you what’s most urgent based on pending quotes, outstanding claims, or market trends. That’s where we see things heading.”