Built around the moments that matter most in mortgage servicing — application status and doc classification, rate reviews, offset and redraw, discharge, and NCCP-safe hardship handovers. The agent doesn't just answer — it takes action, scoped product-by-product, guardrailed for regulated finance.
Loaded with a synthetic dataset covering an in-flight application (Sarah Nguyen, APP-4823) and a settled customer (Michael Chen, HL-2001, Straight Up variable at 6.14%).
Click any prompt to copy it, then paste into chat.
michael.chen@example.com (existing Straight Up variable, 6.14%, 18 months on rate) or reference APP-4823 (Sarah's in-progress application, missing a bank statement). The agent gracefully steers back to these if you throw made-up IDs at it.
Same agent, same workflows, same brand voice — on a real Sydney line. Call the number to try any of the six inbound scenarios — application status, doc submission, rate check, offset, discharge, hardship. The agent asks one question at a time on voice, spells back names and emails before any write, and hands off to Athena Assist on any hardship signal.
Six topic-matched workflows plus two outbound scenarios. Each one asks less, does more, and hands off cleanly when a specialist should own it.
Reads status, names the specific missing doc, and accepts a payslip / bank statement / ID / rates notice / certificate of currency inline. Classifies, validates, and routes to the credit officer with a same-business-day pickup.
Quotes the current rate + time on rate, then lodges a repricing request without predicting the outcome. Specialist calls back within 2 business days with what Athena can offer — no invented percentages.
Answers offset balance, redraw available, next direct debit, features on the loan. Explains fixed-vs-variable mechanics honestly (offset is paused during fixed periods) and never invents fees.
Confirms the loan, reason (sale / refinance / payout), and target settlement date, quotes back before firing the write. Lodges the discharge request and lets the specialist confirm the exact payout — never a made-up figure.
Any signal of financial distress (job loss, missed payment, separation, medical costs) — the agent does not advise. It acknowledges once with dignity, collects name + best number + a one-line summary, and hands off to Athena Assist with priority.
STEER guardrails on every reply: no personal financial advice, no approval predictions, no invented rates or Athena Assist criteria. Crisis signals get Lifeline 13 11 14 and a human, never mortgage mechanics. Off-product asks steer back to Athena.
Two live outbound scenarios you can trigger to your own mobile in our session. Same agent, same guardrails, calling out instead of picking up.
Grounded in Athena's real product taxonomy and the two use cases the founders flagged as highest-leverage.
Scraped athena.com.au's FAQ pages, hardship-assist, eligibility, and target-market so answers are grounded in Athena's own policy language — not invented.
Six inbound topic workflows plus two outbound scenarios. Each one asks one question at a time on voice, quotes back before any write, and hands off cleanly for hardship.
Five mock tools stand in for the Salesforce / Q2 / Front / Vonage stack — application lookup, doc submit, loan account read, repricing lodge, discharge lodge. In production these are your real APIs.
NCCP-safe guardrails (no advice, no approval predictions, no invented rates), a crisis check with Lifeline 13 11 14, and a hardship-first rule that overrides every other flow.
A path from demo to production, phased so risk stays bounded at each step.
Multi-region deployments — Australian customer data can stay in AU. NCCP + Privacy Act aligned by default.
Straight Up, Liberate, Power Up, Athena Select — each product's boundaries are scoped separately, and hardship overrides every other flow.
Escalation isn't a dead end. Hardship, complaints, advice-seeking, and complex broker cases route to a specialist with the customer's context intact.
Regulated consumer credit is where Lorikeet does its best work. Carmoola took end-to-end resolution from 30% to 60% on auto lending.
Lorikeet fits alongside your current stack — inbound Front tickets, Salesforce Service Cloud CRM, Vonage voice — without a rip-and-replace.
As Athena shifts foundations to AWS Bedrock, Lorikeet plugs into the same ecosystem — clean interfaces, no vendor lock-in on the model layer.
Let's walk through how this connects to Front + Salesforce + Vonage, and what the first 90 days would look like.
Talk to the team