What it is
Lorikeet is an autonomous AI support agent built for complex cases rather than simple questions.
What you'd use it for
You would use Lorikeet for support that takes several careful steps: a delayed order, replacing a compromised card, an identity check. The agent authenticates the customer, checks back-end systems, performs the action, and reports back.
Examples of use
A customer reports their payment card has been compromised. The AI agent works through the replacement step by step, running each check in the right order before it acts.
A caller asks for a sensitive change to their account. The agent runs an identity check first, authenticating the customer before it goes any further.
A shopper's order is delayed and the answer sits in several different back-end systems. The agent checks each one, performs the action needed, and reports back.
A fintech customer wants to know the status of a transaction. The agent reads the live data and answers, rather than pointing them to a help article.
A regulated workflow has a fixed procedure that cannot be skipped. The agent follows it exactly, working the case as a strict step-by-step graph.
A case is too involved for a simple FAQ bot to settle. The agent breaks it into tightly scoped steps and works it the way a human agent would.
How it works
Lorikeet uses what it calls an intelligent graph. Instead of free-form reasoning, it breaks each task into tightly scoped, deterministic steps, laid out as a graph that mirrors how a human agent works a case.
How it compares
Where Sierra, Decagon, and Ada aim at broad ticket volume, Lorikeet's pitch is the hard, high-stakes cases. Its strict step-by-step graph is the trade: less open-ended than its rivals, more predictable on complex work.
What others say
No Gartner or Forrester evaluation, which is expected for a company at this stage. No credible independent analyst assessment exists yet.