Pictured: Angus McCullagh, Chief Executive of EML Solutions
When a worker gets injured, the clock starts ticking on a process that has, for decades, been defined as much by paperwork as by people. EML Group is betting that won’t be the case in the future.
Australia's largest specialist personal injury provider recently marked 115 years in business, a milestone that chief executive of EML Solutions Angus McCullagh is using not as an occasion for nostalgia but as a launching pad for the future. EML is pursuing what may be the most ambitious technology agenda in Australian personal injury claims management: the elimination of administrative burden from case management entirely, within three years.
"We want to have bold goals, like Admin-free case management to make sure our organisation's pointing in the right direction," said McCullagh. He sees the current moment as a fundamental transformation for the sector, a sea-change that will clearly demarcate a ‘before’ and ‘after’.
McCullagh argues that previous waves of automation, built on rules-based logic and conditional statements, could never fully address the complexity of personal injury claims. AI changes that calculus.
"You've had phases of automation where you can try and automate things away with a lot of ‘if-then’ statements. But AI makes it actually possible to pick up some of these more complex use cases. Personal injury is not straightforward most of the time, so we haven’t been able to automate a lot of it in the past," he said.
At 115 years, the Insurance Business Australia award-winning provider is well placed to take that step. As a mutual, it has both the remit and the resources to reinvest in better outcomes for its members rather than waiting for others to lead.
"We're at a serious fork in the road. There's no need for our sector in Australia to wait until the world's done it or other people in Australia to have done it," he said.
One number animates the entire strategy: roughly 70% of a case manager's time is currently spent on administration. Drafting letters, typing up call notes, reviewing documents, logging interactions. The remaining 30% is where the valuable work happens: building rapport with an injured worker, motivating a return to employment, coordinating between employers, treatment providers, and specialists.
McCullagh argues the industry has long had this ratio backwards.
"The 30% is what drives the outcome, not the 70%. We should see a massive change in results as we are able to implement more and more technology, because case managers will be able to spend more time getting on the phones and speaking with the stakeholders who help us deliver the outcomes for the individuals who are injured," he said.
The shift he is describing is already underway. EML has deployed AI-generated call transcription and summarisation, meaning case managers are no longer typing notes while simultaneously trying to hold a meaningful conversation. When McCullagh first trialled the tool with one of EML's teams, the reaction from frontline staff caught him off guard.
"I was so excited about the time saving, and they were so excited about the fact that they could have meaningful conversations with workers and employers and doctors where they weren't focused on trying to write the notes. They were focused on actively listening and responding in a way that was driving towards an outcome instead of making it transactional," he said.
That distinction, between transactional contact and genuine engagement, sits at the centre of what EML is trying to achieve. In personal injury, the path back to work is rarely determined by the injury itself. McCullagh is direct on this point: for the majority of claimants who remain off work beyond a few weeks, the main reasons are relational and circumstantial.
"The injury itself, for the majority of claimants, is not the primary driver of a return to work outcome. It's their relationship with their employer, their relationship with their boss, their caring duties at home, how far they are travelling to treatment or to their workplace. All kinds of things that don't appear in structured system fields," he said.
This is the insight behind EML's investment in AI that works on unstructured data: call transcripts, file notes, medical correspondence. Claims files in workers compensation can run to thousands of pages. The ability to surface a buried one line reference to a pre-existing knee injury, or to flag that a worker has repeatedly expressed reluctance to return to a particular employer, changes the entire direction of how a claim is managed.
"We can now identify, using unstructured data, which claims have barriers on them that are not the person's injury itself, which claims have people who have said they don't want to go back to work there, yet the structured data on that claim still says our goal is to return them to the pre-injury employer in their pre-injury role." McCullagh explained.
What distinguishes EML's approach from off-the-shelf technology deployments is the decision to build its AI capability entirely in-house, with frontline staff embedded in the design process. The organisation has been building its own technology for 30 years, with no external vendors involved in its core systems. It has recently invested in purpose-built AI infrastructure, including millions of dollars' worth of dedicated servers that keep sensitive claimant data entirely within EML's own network.
That infrastructure choice is partly a security decision. Workers’ compensation files contain sensitive personal information: health records, home addresses, financial circumstances, family situations. McCullagh is unequivocal about the risk of routing that data through external AI services.
"We know that we are secure. I wouldn't want my [personal] information put into a chatbot. So why should we put an individuals information into something that we don’t control and therefore can’t absolutely guarantee its security?" he said.
The in-house model also means EML's AI team, which includes several professors in the field employed full-time, can innovate without watching a cloud computing bill climb in real time.
Critically, the case managers themselves shape what gets built.
"Technology cannot be disconnected from process these days. You can't say let's improve the process [and do it] with poor technology. Our people control the technology, our people control the process. Case managers sit with the technology business analysts and say this is how I want to do this process, because I know how workers will respond to that," he said.
EML is also careful about where AI acts and where humans remain firmly in the loop. Every AI-assisted process retains human review. A drafted letter is a prompt for a case manager's professional judgement, not a finished product.
"There's a lot of nuance in claims. We don't have any vision to remove the human from the AI flow. The AI is really just there to support, remove the administrative burden, and give you information to say here's a draft, what do you want to do with it? Then our claims teams bring their professionalism and experience to finish off that draft, customising it as needed,” McCullagh said.
McCullagh is insistent that technology should sharpen what case managers do, not replace them. The basics of good claims management, prompt communication, proactive outreach, genuine human connection, will matter more in an AI-assisted environment, not less.
"We have to make sure that the human element is what comes to the fore even more than it is today. That's what gets claims outcomes today. But it will get even better claims outcomes when we can free up administration time and let case managers spend that time picking up the phone and building rapport with someone," he said.
In New South Wales, where EML has its largest workforce, legislative reform in the workers compensation system is adding complexity to an already demanding operating environment. EML has been an active participant in parliamentary consultations around the Workers Compensation Legislation Amendment Bill 2025. It is building workflow technology specifically designed to guide case managers through the decision-making requirements of the new legislation, including a large number of decision checkpoints for assessing psychological claim liability.
"When you have a lot of checkpoints to work through, and you rely soley on humans to do that, they will make errors. We have to be able to help them do that through technology," he said.
For self-insurers and specialised schemes operating within the NSW system, EML's combination of scale, technological depth, and regulatory engagement positions it as the partner of choice through the transition.
Many organisations would be intent on preserving their incumbency at 115 years old. For EML, the milestone is a checkpoint it is passing before it accelerates into a period of change that will shape the sector for the next century.
"I do believe that all claims management, but especially personal injury claims management, is going to fundamentally change. The impact of AI and technology will change what the priorities are in claims management. And that doesn't diminish the people; it heightens the people. It makes it less about administration and time management and more about conversations, relationships, rapport and influence to help deliver an outcome on claims," he said.
This article was produced in partnership with EML Group
