Onpode
Cover art for Suncorp deploys AI agents in claims this month; Lendi rebuilt for the agentic era

Suncorp deploys AI agents in claims this month; Lendi rebuilt for the agentic era

June 22, 2026 · 5 min

Marcus Vale & Ben Okonkwo

Suncorp Group deployed five live AI agents inside its insurance claims process in June 2026 — handling FNOL capture, classification, routing, vendor dispatch, and settlement calculation — with no published error rates or legally defensible audit trails. Meanwhile, Lendi Group rebuilt its entire mortgage platform from scratch for agentic AI, not retrofitted.

Australian financial institutions Suncorp Group and Lendi Group are among the earliest public deployers of agentic AI in regulated financial services workflows, marking a tangible industry shift from conversational chatbots to autonomous, multi-step task execution systems. Suncorp, a major insurer, is rolling out five AI agents into its insurance claims process by the end of June 2026.

0:005:06
Make your own on Onpode

Describe any topic. Hear it in minutes.

About this episode

Australian financial institutions Suncorp Group and Lendi Group are among the earliest public deployers of agentic AI in regulated financial services workflows, marking a tangible industry shift from conversational chatbots to autonomous, multi-step task execution systems. Suncorp, a major insurer, is rolling out five AI agents into its insurance claims process by the end of June 2026.

Frequently asked

What AI agents is Suncorp deploying in its insurance claims process?

Suncorp Group deployed five AI agents in its insurance claims process in June 2026, handling FNOL capture, claim classification, routing, vendor dispatch, and settlement calculation. Kranthi Nekkalapu, Suncorp's AI practice executive, presented the deployment at the Databricks Data+AI Summit in San Francisco as a live production system, not a proof of concept.

Can an AI agent legally deny an insurance claim?

When a Suncorp AI agent flags a policy exclusion and issues a denial, it is making a legal determination — not just a processing decision. No human checkpoint has been named in Suncorp's public architecture, and no legal framework currently assigns responsibility if such an automated denial is wrong, according to analysis by Ben Glenney at Restive VC.

How did Lendi Group rebuild its mortgage platform for agentic AI?

Lendi Group did not retrofit its existing mortgage platform — CEO Dave Hyman rebuilt it entirely with a proprietary orchestration layer integrating over 30 lenders, then launched monō ai on top of that new foundation. The rebuild reflects a structural conviction that incumbents who only layer agents onto legacy systems are already architecturally behind.

Are AI claims denials by insurers biased or legally defensible?

US insurers are already seeing two-to-three times claims denial spikes attributed to biased AI, with no defensible audit trails, according to data surfaced by Kasey Roh at upstageai. Suncorp has built a Databricks-based observability platform using OpenTelemetry, but observability — logging every agent step — is not the same as a legally defensible audit trail.

What is mount.insure and why does it matter for AI agent liability?

Mount.insure is an emerging insurance carrier purpose-built to underwrite and red-team autonomous AI agent deployments, flagged by YC's fintech partner. Its existence signals that the liability gap created by live agentic AI systems — like Suncorp's claims agents — is real enough that a new insurance product category has formed specifically to cover the agents themselves.

Grounded in 12 sources
What is "fundamental"? · arxiv.org
What Are A.I. Agents Actually Doing? · nytimes.com
Agentic AI And The Next Layer Of The Trade: Beyond Nvidia - Forbes · forbes.com
Google Sued Over AI Search, the Future of Touchscreen MacBooks, and Canada Targets Teen Social Media Use | Tech Today - CNET · cnet.com
Beyond Siri: Here are the practical AI features coming to your iPhone in iOS 27 - TechCrunch · techcrunch.com
OpenAI launches scheduled tasks in ChatGPT, details here - 9to5Mac · 9to5mac.com
AI agents are getting invested in banking - Newsweek · newsweek.com
Suncorp to have AI agents in insurance claims process as soon as this month - iTnews · itnews.com.au
Study: AI is reshaping automotive finance despite scaling challenges - Consultancy.eu · consultancy.eu
Agentic AI in Insurance Claims: Where to Start, What to Automate, and What to Keep Human-Controlled: By Vital Soupel - Finextra Research · finextra.com
How Lendi Group rebuilt its mortgage platform for the agentic era - Poseidon · poseidon-us.com
AI in Fintech: Lendi’s Dave Hyman Teams Up with Danny Gilbert on monō ai (2026) · librettoworld.com
Read transcript

Ben Okonkwo: You look like someone who just read something they can't stop thinking about.

Marcus Vale: Kranthi Nekkalapu, Suncorp's AI practice executive, presented at the Databricks Data+AI Summit in San Francisco. And I — look, I've heard a lot of 'AI in financial services' pitches. This one's different.

Ben Okonkwo: Different how?

Marcus Vale: Five agents. Live. June 2026. Suncorp Group is not running a proof of concept — they are in production in a regulated insurance claims process. And Lendi Group has rebuilt its entire mortgage platform for this. Not retrofitted. Rebuilt.

Ben Okonkwo: Now — I want to make sure listeners actually grasp what agentic means, because it gets thrown around. What's the intuition?

Marcus Vale: Personal assistant versus vending machine. A chatbot gives you one answer to one question. An agentic system takes a goal and executes across multiple systems — makes decisions, handles ambiguity, loops back. No hand-holding.

Ben Okonkwo: And Steve Johnston, Suncorp's CEO, is calling this 'full-scale delivery' — not a trial.

Ben Okonkwo: Right, and I want to pull on that phrase — 'full-scale delivery' is Steve Johnston's characterization. Operationally what Kranthi Nekkalapu presented at the Databricks Data+AI Summit is five agents handling discrete sub-processes. FNOL capture, classification, routing, vendor dispatch, settlement calculation. Those are — I mean, they're real, but they're parallelized workflow automation. That's not the same as an autonomous system making end-to-end decisions.

Marcus Vale: Fair distinction. What's the load-bearing assumption you're worried about?

Ben Okonkwo: Suncorp has published zero data. No error rates, no false denial rates, nothing on how a customer even challenges a decision the agent made. They built a Databricks-based observability platform ingesting OpenTelemetry feeds — which is actually a smart infrastructure signal — but observability is not auditability. You can log every single step an agent takes and still not be able to explain that denial to a regulator in legally defensible language.

Marcus Vale: The logging exists. The legal language doesn't.

Ben Okonkwo: Exactly. And — wait, actually this is where it stops being theoretical. Kasey Roh at upstageai surfaced data showing US insurers are already seeing two-to-three times claims denial spikes attributed to biased AI. No defensible audit trails. That's documented, that's active. So if Suncorp's agents inherit that kind of bias at scale—

Marcus Vale: They're not capturing efficiency. They're capturing lawsuit velocity. And Ben Glenney at Restive VC flagged exactly this — autonomous agents create liability and authority problems before any legal framework exists to assign responsibility. Nobody's been the first defendant yet.

Marcus Vale: So walk through it concretely. Tuesday, 9:47 a.m. Customer files a water damage claim. Suncorp's FNOL agent ingests the report, classifies it, routes it. By 10:04 a.m. — seventeen minutes — a second agent has calculated settlement. Except it flagged 'flood' in the description, matched a policy exclusion, and issued a denial. No human checkpoint named anywhere in the public architecture.

Ben Okonkwo: And that's — wait, that's not a processing error. That's a legal determination.

Marcus Vale: That's what I mean. Suncorp has publicly confirmed they're developing agents that determine whether a claim should be covered at all. That's not efficiency anymore. That's the agent writing the legal decision. And they've named the ambition — nobody's named who's liable when it's wrong.

Ben Okonkwo: Okay and — actually, the Lendi Group piece sharpens this, because Lendi integrating over thirty lenders with a proprietary orchestration layer isn't retrofitting. Dave Hyman didn't layer agents onto the old platform. He rebuilt entirely and then launched monō ai on top of that conviction. That's a structural bet that incumbents who retrofit are — I mean, they're already behind architecturally.

Marcus Vale: Pennymac's doing the same thing in the US — expanded their AWS agreement, built a voice assistant on Amazon Nova Sonic, modernising their Plaisse servicing platform. Different geography, same signal.

Ben Okonkwo: So the first defendant — whoever that is — they're not just losing a lawsuit. They're writing the compliance playbook every other insurer copies.

Ben Okonkwo: And that's — okay, that's the tell for me. Mount.insure. YC's fintech partner flagged it — it's an emerging carrier purpose-built to insure autonomous AI agent deployments. Red-teaming them. Writing coverage for them. Which means... I mean, the insurance industry has created a liability gap so real that a new insurance product exists to fill it. For the agents. The agents now need their own insurance.

Marcus Vale: Fine. The tech works. Suncorp is shipping. I'll give you that.

Ben Okonkwo: The agents are live. The legal frameworks are not.

Suncorp deploys AI agents in claims this month; Lendi rebuilt for the agentic era · Onpode