Lorikeet
for MYOB
Walk the complete customer journey — from in-product chat, through L1 resolution and L2/L3 handover, to HubSpot write-back, AXIS sandbox provisioning, and Coach. Click any prompt below to drop it into the chat and watch the agent work.
The agent embeds as an iframe inside MYOB Acumatica. A single public key maps to MYOB at the subscriber level — instance-agnostic, so 5,000 tenants is the same deployment as one. Each conversation arrives with screen ID, URL, and current user packed as a JSON blob and exposed to every workflow as custom attributes — so the agent qualifies fast and doesn't re-ask what the customer's screen already told it.
The agent triages each ticket against the live workflow library. It either resolves end-to-end at L1, applies partner-appropriate language for partner-submitted tickets, or routes to a human at L2 / L3 — with the full diagnostic payload, known-issue match, and an AXIS sandbox already provisioned. Click any prompt to test live.
Every ticket resolves in-flow against the MYOB Acumatica help centre (scraped end-to-end). The agent calls search_knowledge first on every turn, quotes the article verbatim for any regulatory rate or threshold, and never paraphrases procedural steps from memory.
Pay processing errors
Auth & notifications
Tax & compliance (AU/NZ)
Pay Super
WFM integration
Terminations
Leave management
Partner channel skips end-customer ID verification (the partner is already authenticated). The agent uses technical, partner-appropriate language — no consumer hedging or "have you tried turning it off and on again". Today these land in L2 by default; the agent identifies the ones an L1 workflow can actually resolve and handles them inline.
Tickets that contain a verbatim Acumatica error, match a known integration bug, or need per-case AP/AR scripting route to a specialist queue. The agent collects every diagnostic field (only those not already in widget metadata), runs checkKnownIssueByErrorMessage against the ADV ticket database, auto-orders an AXIS sandbox where one's needed, and routes with the complete investigation payload. The engineer picks up an environment that's already ready.
Native HubSpot integration. At the end of every conversation — resolved or escalated — the agent emits a structured engagement against the customer's contact and company. The exact payload below. The full conversation transcript and call metadata attach automatically. Your team continues to work entirely in HubSpot.
Try any pill above and the agent will emit this payload before closing the ticket. Internal notes also post to HubSpot mid-conversation when the agent does notable diagnostics.
For every L2 / L3 routed in Stage 2, the AXIS sandbox is provisioned the moment the agent decides to escalate. The engineer who picks up the queue inherits the structured diagnostic payload, the known-issue match (if any), the tenant build, the customisation pack version, and a live sandbox already cloned to the customer's configuration — no setup overhead before they can start investigating. For known issues with a documented fix script, the engineer moves straight to applying it; for novel issues, they start one step ahead instead of one step behind.
Coach reviews every ticket against your scorecard and topic taxonomy. Where the agent escalated something it shouldn't have — or resolved something it shouldn't have — Coach surfaces it. Non-technical users ask Coach in plain English ("what drove the spike in pay processing tickets this week?") via the Lorikeet UI or MCP, and get the topic breakdown, top failing tickets, and a proposed KB article or workflow change. Coach can draft the article and queue it for review — your team approves, it goes live, coverage climbs.