Lorikeet
for Real Insurance
Contact us
for Real Insurance
for Real Insurance
Lorikeet is an AI customer support agent built for regulated industries like insurance. This demo runs on Real Insurance's real product portfolio. The same agent handles outbound Quick Quote callbacks, inbound voice, and inbound chat — with a clean handoff to a human specialist whenever judgement is needed. Pick a customer below and the agent will ring your phone within seconds.
Pick a customer scenario, enter your mobile, and the AI agent will ring you to follow up on their Quick Quote.
The same agent that runs the outbound callback also handles inbound questions on chat or voice — existing-policy servicing, claims intake, FAQs, and the conversational version of the Quick Quote form. Click any prompt below to drop it into the chat widget, or ring the voice line.
Click any prompt to copy it.
Type anything you like — the agent rolls with whatever name, policy, or details you give it.
Three categories, mapped to Real Insurance's existing Genesys queues. The agent confirms details before any write and routes anything that needs human judgement back to the right specialist with full context.
The agent runs the Quick Quote intake conversationally on inbound voice or chat — same fields, same order as your web form. On outbound, it calls customers who didn't complete sign-up, recaps the exact cover amount they picked, handles standard objections, and warm-transfers to a Sales specialist in Genesys.
The everyday admin that doesn't need a specialist — updating contact details, swapping a card or bank account, resending an annual statement or policy schedule, or rescheduling the next premium debit within the 14-day policy rule. Spell-confirmed, masked where it matters, written back across every active policy in one tool call.
Anything that needs human judgement — a new claim, beneficiary change (signed form), cancellation, or a customer in distress — gets a calm, structured intake then a warm transfer to the right Genesys queue (Sales, Service, or Claims) with full context. The agent never makes a claim decision, quotes a payout, or pretends to action a regulated change live.
Four steps. Grounded in Real Insurance's public website and the conversations a regulated-insurance support team has every day.
Scraped realinsurance.com.au and walked through every Quick Quote form (Life, Funeral, Income Protection) on the live site. Each agent reply is grounded in Real Insurance's actual products, policies, and contact-centre queue structure.
Nine plain-English conversation flows — one per topic — plus a knowledge-base fallback. Each one repeats key details back to the customer before saving anything, and follows Real Insurance's product structure.
Eleven mock backend calls stand in for the real integrations — looking up a customer, writing a contact-detail change, transferring to a Genesys queue, capturing a claim, logging an outbound call outcome. The Quick Quote payload is real enough that the outbound recap names the actual cover amount the customer picked.
Sixteen conversation scenarios — outbound sales, inbound servicing, claims, beneficiary change, cancellation, FAQ, and a crisis path — re-run five times each. The same suite runs every time the conversation flow changes, so regressions get caught before any real customer hears them.
Where this would go in production — same agent, plugged into Real Insurance's real systems.
Replace the mock backend calls with real integrations — the policy admin platform, Genesys Cloud queues, and the Quick Quote form pipeline. Read-only access first, so the agent can look things up before it can change anything.
Run the outbound Quick Quote callback on real customer records. Auto-tagged transcripts and conversion outcomes feed Power BI directly — no per-call manual review.
Switch on the inbound self-service flows (contact details, direct debit, document resend, premium reschedule). Every write is gated by a spell-confirm and a guardrail check. Beneficiary, claims, and cancellations stay warm-handoff to a human specialist.
Same platform, same simulation suite, same reporting — extended to Australian Seniors, Buddy, Choosi, Guardian, Kogan, and the Canada expansion. One change tested across the whole portfolio in minutes.
Lorikeet was built for regulated industries — insurance, financial services, healthcare — where privacy, residency, and conversational safety are non-negotiable.
Lorikeet runs in Australian, American, and European data regions. Australian customer data can stay in Australia. The same agent setup serves the Canada expansion from the American region.
A guardrail blocks the agent from recommending a cover amount, comparing competitors on suitability, or steering customers to a specific product. Every advice question routes warmly to a human specialist.
If a customer shows signs of distress, the agent surfaces the right local crisis support resources and routes immediately to a human team member, skipping any routine flow.
Every change to the agent's behaviour automatically re-runs sixteen conversation scenarios. Regressions are caught before any real customer hears the new version — no need to pause production to fix a prompt.
Every conversation ends with a structured outcome — transferred, callback, declined, objection categories. Power BI and Datadog get clean events. Quality reviewers focus on exceptions instead of listening to every call.
Lorikeet drops the conversation into the right Genesys queue (Sales, Service, or Claims, by product) with structured context. The human specialist picks up where the agent left off — the customer doesn't repeat themselves.
Happy to walk through the simulation suite, the Genesys integration, and how this would land in production.
Talk to the team