Insurance · Sydney
Claims faster. Conduct cleaner. Underwriters happier.
We help Australian general, life and health insurers ship AI that survives APRA prudential review, LIF design-and-distribution obligations, and the next Royal Commission moment. Claims automation, conduct surveillance, fraud detection and underwriting AI — built with auditors in the room.
Get a Sydney-based response within 1 business day
Why insurers call us
The work where the loss ratio and the regulator both matter.
Insurance is one of the highest-leverage industries for AI in Australia — and one of the most heavily-regulated. We build AI workloads that move the loss ratio and clear APRA review at the same time.
Our travel claims volume tripled. Adjuster headcount didn't. Touch-time per claim is killing NPS.
Adviser conduct review is one person sampling 2% of calls. We just had a breach we should have caught at month 3, not month 14.
Our underwriting AI works in pilot, but legal won't let us put it into production until 'someone' figures out the bias audit.
Fraud detection is rules-based and 8 years old. False positives are higher than true positives. The fraud team is drowning.
Board asked for 'AI in claims by FY end'. We don't have a vendor list, an architecture, or a person to own it.
What we ship
Five insurance AI workloads we've actually shipped.
Claims Automation (Travel / Motor / Home)
Agentic claims triage with document understanding, policy-cover matching, fraud signals and auto-resolution for low-complexity claims. Reference: travel insurance claims automation in AU/NZ.
Adviser Conduct Risk Monitoring
Call-recording surveillance with anomaly detection, hawking-language flags and audit-grade evidence. Reference: conduct risk monitoring with Claude + D23.io.
Underwriting AI with Bias Audits
Decisioning models with explainability, drift monitoring, and statutory bias audits so production sign-off survives APRA and consumer-credit challenges.
Fraud Detection (Next-Generation)
Move from rules-based fraud detection to graph + ML hybrid models. Lower false-positive rate, higher recall, with explainability for SIU and human-in-loop review.
Embedded Analytics on Apache Superset
Replace per-seat BI (Tableau / Power BI) with embedded Superset + ClickHouse for actuarial, claims, distribution and exec dashboards. AU data-residency by design.
What "good" looks like 90 days in.
Reduction in claims-handler touch-time on low-complexity travel and motor claims with agentic triage.
Adviser-call coverage on conduct surveillance vs the typical 2-5% manual sample.
Improvement in fraud true-positive rate when moving from rules-only to graph+ML hybrid detection.
APRA prudential and ASIC supervisor evidence posture, signed off without remediation.
Stop benchmarking. Book a call.
Thirty minutes. We'll either tell you what to do next, or who you should be talking to instead.
Book a 30-min advisory call