Lorikeet
for MYOB
Contact us
Three scenarios. Three lanes. One shared auth gate. Each step labelled with who does it and which system it touches. The companion to the Acumatica user journey demo — that one shows the agent working live, this one shows what your team's work day looks like with it in.
Per Patrick (4 Jun): for Direct Pipeline, ID verification must be based on the authorised user flag on the contact record in HubSpot against the correct company record — not the contact name on the account. Compliance breach occurs when account-specific information is disclosed to an unauthorised recipient. Lorikeet enforces this as a structural gate. The tool can't fire unless the gate passes.
authorised_user = true. Check contact's company association matches the company on the account. This mirrors MYOB's current ID verification standard for Direct Pipeline.Customer or partner asks how to do something in Acumatica. The agent triages whether the question is account-specific or general, runs the auth gate if needed, then searches the live KB and quotes the article verbatim for any regulatory rate or threshold. Today: human pickup, lookup, paraphrase, reply.
Patrick called this out as the canonical "L1 task customers can't self-serve, but a human team completes today". The current process is 8 steps via Slack + Stargate Dialer. With Lorikeet: auth gate → reset_mfa tool → confirm → close. Human only invoked on auth failure.
Either works. Decision is yours; we scope to fit.
L2 is where Lorikeet's value compounds. The agent collects the full diagnostic payload, matches against known issues, provisions the AXIS sandbox the moment it decides to escalate, and writes a structured HubSpot ticket. The engineer starts one step ahead instead of one step behind.
Both phases include forward-deployed engineering and product, all integration work, and ongoing iteration. There are no setup fees, no implementation fees, and no per-seat charges — everything is wrapped into the annual commitment.
The two highest-volume, lowest-complexity flows. Get the agent in production, in front of your customer and partner base, and return clear value before adding L2 scope.
Once Phase 1 is stable and showing clear value, layer in L2. Lorikeet collects the diagnostic payload, matches against known issues, provisions the AXIS sandbox, and hands the engineer a sandbox-ready ticket.
Lorikeet charges per successful resolution. A resolution is the AI agent completing the workflow as designed — including paths where the designed outcome is a warm handoff to a human (the agent has done L1 qualification and routed with full context). You don't pay when the agent fails, confuses the customer, or your team marks the ticket bad.
AI completes the designed workflow. Includes L1 triage that ends in a structured handoff to L2 / L3 — the agent still did the work to qualify, capture diagnostics, and route correctly.
Does not include: failed tool calls, agent off-script, customer-confused outcomes, or tickets your team marks bad.
The $48k / $90k figures align to indicative Scale Plan rates. Higher annual commitments unlock lower per-resolution rates.
If actual volume runs materially above the resolution band, we work with you on a tier adjustment — the economics scale, the per-unit cost comes down.
Lorikeet's forward-deployed model means we do the build — but the agent only gets to "MYOB-grade" if your team puts time into shaping workflows, transforming SOPs that don't live in docs, and rigorously testing every workflow before it goes live. Front-loaded at the start, much lighter at steady state. The workflow testing effort is the single biggest line item — it's what gets us from "looks right" to "actually right for MYOB customers".
First 6–8 weeks (build): face-to-face heavy — weekly working sessions with FDE/FDPM, knowledge transformation, intensive workflow testing.
Weeks 8–12 (stabilisation): daily monitoring during first 2 weeks live, then weekly — Coach insight review, edge-case patches, performance tuning.
Steady state: fortnightly review with FDPM, ~2–3 hours / week from ops to handle new edge cases. Lighter contact, more strategic.
Today, only customers and partners with a real issue raise tickets — that's the ~4,000 / year baseline behind Phase 1's commitment. With the agent embedded in Acumatica and 10,000+ tenants able to chat freely (no friction barrier), traffic may run materially higher than that.