My summary of Lorikeet before looking into it – An AI powered customer support agent.

How does Lorikeet explain itself in the first minute? “The AI Customer Concierge for complex companies”, which to me suggest that Lorikeet aims to provide a solution to dealing effectively with more complex support queries.

Examples such as “flight rescheduled” and “medication express shipped” are listed on the Lorikeet homepage. “Where other agents deflect, Lorikeet leans in, handling high-stakes problems with precision” is the promise behind the product. Lorikeet can integrate with the customer support systems teams are already using. It treats products such as Zendesk, Hubspot, Sierra and Intercom as ticketing systems. Lorikeet’s concierge agent can then monitor the tickets and respond to them. Combining natural-language capabilities with deterministic, step-by-step workflows is how Lorikeet positions its USPs, specialising in highly regulated industries.

How does Lorikeet work? Lorikeet has two main systems it uses serve customer experiences: an AI Concierge and a Coach. The Concierge is positioned as a proactive agent, that won’t just act on support tickets but will do things like following up on abandoned shopping carts and serving product features based on usage behaviours. The idea is that you can use Concierge to manage the entire relationship a customer has with your business, from sales to customer support.

To enable this, the Concierge needs cross-system memory (not just ticket history), multi-channel capability (not just chat), outbound as well as inbound interaction, and the ability to operate across different business functions with different rules and objectives. Business can self-configure the Concierge, by using their agentic platform of choice (e.g. Claude, Cursor or ChatGPT).

“Workflows” in Lorikeet is the set of instructions and tools that tells an AI agent how to handle a customer query. When an agent searches for “workflow,” it gets all workflow-related tools, when it searches for “tickets” it finds all ticket related tools.

Finally, there’s Coach, a Lorikeet agent that monitors the end to end customer experience across Lorikeet but also Slack, Claude and ChatGPT. Using predefined skills, Coach can monitor and improve specific workflows, suggestions that you can implement or that Coach can implement on your behalf.

Main learning point: Lorikeet has made two clear strategic choices: go deep on highly regulated industries like fintech and insurance, and work with existing support infrastructure rather than against it. Whether that combination, delivered through its Concierge and Coach agents, is enough to stand out in an increasingly crowded AI support will be interesting to watch.

Related links for further learning:

1. https://www.lorikeetcx.ai/blog/lorikeet-is-now-self-configuring-and-agent-first
2. https://www.linkedin.com/posts/michaelmomi_we-just-shipped-the-official-lorikeet-mcp-ugcPost-7445698157507842048-rAVz/
3. https://www.lorikeetcx.ai/blog/how-we-built-lorikeet-mcp-to-be-safe-and-powerful
4. https://medium.com/aimonks/what-are-agentic-tools-f5bef3134100