Childcare support that acts, not just answers.
An AI agent built for Koru Kids — checks the family invoice against the Koru shift log, raises a PIQ with Payroll when the numbers don't match, and proactively re-engages nannies whose placements have ended. The complex cases your CX team currently spends hours on.
Try it on the phone
Tap the chat bubble inside the Koru Family Hub on the left, then paste any of these. The agent reads the message, calls the right tool, and replies in seconds.
Not a deflection bot. An action-taking agent.
Lorikeet doesn't just match the message against an FAQ. It reads the ticket, calls the right tools across Xero, the Koru platform, and Airtable, reasons over the data, and drafts the action — or escalates with full audit trail.
Multi-system context, natively
Pulls the family's Xero invoice line, the Koru shift schedule, and the nanny's leave log — in parallel — and reconciles them. Then raises the PIQ in Airtable and drafts the family + nanny replies for human review.
Two personas, one agent
Distinguishes family vs nanny conversations and pulls the right context for each. Family-side payroll queries; nanny-side application, DBS, and leave questions — same agent, same workflows.
Proactive outbound, not just inbound
When a placement ends, the agent texts the nanny to re-list — captures their availability, updates the Koru platform, logs Airtable. The growth motion you're already running with humans, scaled.
Built in days, not quarters
Zero custom code. Every workflow, tool, and guardrail configured through the Lorikeet product. Same shape as your real backend — Koru platform, Xero, Airtable, WhatsApp.
Mocked your stack
Tools for the Koru platform (match, shift schedule, leave log, nanny status), Xero (invoice lines), and Airtable (PIQ + re-engagement) — stubbed with realistic data. Same call shape as the real APIs.
Scraped your help centre
60+ articles from intercom.help/korukids pulled into the knowledge base — DBS, contracted hours, notice periods, advance payments, employment contracts. The agent grounds its answers in the same content your team already wrote.
Wrote the workflows
Plain-English instructions tell the agent how to handle a disputed shift, when to raise a PIQ, when to escalate, and how to run the nanny re-engagement SMS flow. No rules engine, no scripted decision trees.
What's inside this demo
The Lorikeet sandbox is open for the Koru team to inspect. Everything you see here is live and editable.
NL workflows
PIQ — Disputed Shift; Outbound — Nanny Re-engagement. Both live and verified end-to-end.
Mocked tools
koru_platform (one tool, all data points), xero, airtable, plus an orchestrator that wires them together.
Knowledge articles
Scraped from your live Intercom help centre — DBS, employment contracts, advance payments, notice periods, application FAQs.
Reply latency
End-to-end on chat, including tool execution, multi-system reconciliation, and reasoning. Same benchmark for production.
Ready to see this on real Koru tickets?
This is a synthetic slice with mocked tools. Connect Xero, the Koru platform, and Airtable and we'll have a pilot answering live tickets within two to four weeks — well before your November Front churn.
Talk to Lorikeet