Agentic AI is now doing work that looks a lot like broking, including fine-print explanations and even end-to-end sales. A Melbourne-based insurance technology company recently announced what it called “Australia’s first fully autonomous, compliance cleared health insurance sale,” handled by an agentic AI platform without human intervention. Konkrd said it also plans to launch digital broking in travel, pet and car insurance this year.
The bigger story for brokers and insurers is that this new wave of AI is no longer just chatbots answering simple questions, or helping staff draft emails. It is starting to resemble a production-grade workflow that can respond intelligently to client needs, interpret product documents, recommend options and transact at scale.
Autonomous brokers may not take broker jobs in one clean sweep. But they are unbundling the role - forcing the industry to decide which parts of broking are truly human craft, and which parts were always destined to become software.
Eric Lowenstein said the Konkrd milestone has signficance, even for intermediaries who do not touch retail health.
“What this signals is the pace of change - autonomous, end-to-end transactions in regulated environments are moving from concept to reality very quickly,” said Lowenstein, CEO of Sydney-based healthcare and medical indemnity underwriting agency Tego.
He said similar initiatives are emerging overseas. UK-based Jointly AI, for example, has also announced the launch of an autonomous brokerage platform.
“The true test will be just how autonomous these systems can become without a human in the loop - though it's worth remembering that automated quoting and purchasing in regulated environments has in one form or another been happening for years through comparison and direct quote websites,” Lowenstein said.
For intermediated-market stakeholders, the core question is shifting. It is less about whether AI “replaces brokers” and more about what these systems can do that prior waves of digitisation could not.
“The potential value is in doing something those [comparison] sites don't - genuinely reading fine print, explaining trade-offs in plain language and tailoring recommendations to a person's actual needs,” he said.
In other words, if “autonomous broking” simply becomes a faster comparison site, it may be useful - but not transformative. The competitive difference, Lowenstein suggested, will be whether AI improves the decision itself.
“The platforms that succeed will be those that can demonstrate they're meaningfully improving the decision, not just the process,” he said.
For digital brokers to be credible in insurance, they must show how compliance and disclosure are handled in real time, in a way that stands up to financial services regulation. The second obstacle is that - even for reasonably straightforward cover - there is still often a need for a human in the loop.
Konkrd’s co-founder and executive chair Scott Wilson said his platform is designed to hard-wire disclosure obligations into the customer journey.
“Disclosure is built into the flow of the needs based funnel,” he said.
The platform is built to ask the right questions in the right sequence, he said, present key product information clearly and maintain a record of what the customer was shown, what was explained, and what they confirmed.
Customers are guided through the policy’s material terms - key exclusions, limits, waiting periods and other important conditions - before they purchase. Plain-English explanations sit alongside the insurer’s formal documents, said Wilson, so customers aren’t simply handed a PDS; they are helped to understand it.
“Just as importantly,” he said. “The platform creates an audit trail of the journey, including the questions asked, the information provided, the disclosures presented and the customer’s confirmations.”
The second pressure point is how often humans still intervene.
“At the moment, around 50% of customer interactions result in a human stepping in at some point,” Wilson said.
So, the industry is not yet in a “zero humans” world. Wilson said there are two different types of human involvement with his firm’s offering: a human joining the conversation when a customer wants it or the case is complex, and a separate layer of supervision and monitoring for compliance and quality while the platform scales.
Lowenstein said health insurance is a logical first proving ground for these types of digital brokers because it is complex for consumers but still relatively standardised compared with bespoke commercial placements.
“This makes it well suited to an AI-driven workflow,” he said.
The fear among brokers - and many information-rich professions - is job displacement. Some of that pressure is already being felt at entry level. But core broker functions, including the relationship-driven, face-to-face elements of the job and (for now) more complex risks, appear less immediately exposed.
“What’s most under threat isn’t broker judgment, it’s the admin layer: chasing quotes, rekeying the same data, comparing long documents, and doing endless back and forth for straightforward risks,” Wilson said.
That aligns with what is already happening globally, even where solutions are not fully “autonomous brokers”. Major intermediaries and vendors are building AI into placement and servicing workflows: Aon has launched “Aon Broker Copilot” to modernise commercial placement using AI and analytics, and UK specialty broker McGill and Partners has announced an “AI agent” built on Salesforce’s Agentforce. In personal lines distribution support, Quandri has positioned AI around automating parts of the renewal cycle, including policy analysis, requoting and client communication.
The more significant challenge for human brokers will come if digital brokers move beyond being bots that can sell.
“In my view, the most compelling differentiator would be an AI system that goes further - autonomously asking smarter questions to build a more accurate picture of someone's risk profile and using that intelligence to adjust pricing and tailor cover to their unique circumstances,” Lowenstein said.
That is where AI could move well beyond what comparison sites already offer today.
“The more interesting question is what happens as this technology moves beyond retail into more complex insurance needs, where products are less standardised, underwriting is more bespoke, and underwriting judgment carries more weight,” he said.
If AI-led distribution improves the quality of complex risk information - through better fact-finds and clearer explanation of trade-offs - that is when the stakes rise for underwriters, insurers and brokers alike.
