Lorikeet MCP Server

Lorikeet is an AI customer support platform that builds "universal concierges" capable of resolving complex, multi-step support tickets across chat, email, voice, SMS, and WhatsApp. Rather than acting as a scripted chatbot, Lorikeet's agents authenticate users, look up data in backend systems, perform actions like processing refunds or updating accounts, and communicate the result back to the customer. The product is targeted at fintechs, healthtechs, and other companies with regulated or complex workflows.

Lorikeet integrates with ticketing systems (Zendesk, Intercom, Salesforce, Front, HubSpot, Help Scout), telephony (Twilio, Genesys Cloud, Amazon Connect), knowledge bases (Confluence, Notion, Google Docs, Guru), and tools and actions providers (Stripe, Shopify, SendGrid). According to Lorikeet's integrations page, MCP servers are one of three supported mechanisms for extending agent capabilities, alongside REST APIs and webhooks. This means a Lorikeet agent can call out to external MCP servers to read data or take action when resolving a ticket.

Note: Lorikeet's primary use of MCP is as an MCP client that connects out to external MCP servers from inside its agent runtime. As of this writing there is no public, self-serve MCP server endpoint hosted by Lorikeet that third-party AI clients (Claude Desktop, Cursor, etc.) can point at. Lorikeet's developer documentation at docs.lorikeetcx.ai is access-restricted, so configuration specifics require a customer account.

Lorikeet supports MCP as part of its "Tools and Actions" configuration system, where MCP servers can be wired into a Lorikeet agent to extend its capabilities. The configuration is managed inside the Lorikeet workspace rather than via a public config file.

Prerequisites

• A Lorikeet account (contact sales at lorikeetcx.ai for access)
• The URL and credentials for the MCP server you want Lorikeet to call
• Access to the customer-only docs at https://docs.lorikeetcx.ai/

General flow (per Lorikeet's public integrations page)

• Log in to your Lorikeet workspace
• Open the Tools and Actions section
• Add a new tool and choose MCP server as the connector type
• Provide the MCP server URL and any required auth (API key, OAuth token)
• Attach the tool to the agent workflow that should be able to call it

Because the configuration UI and exact field names live behind Lorikeet's access-restricted documentation, refer to the official quickstart at https://docs.lorikeetcx.ai/guides/quickstart once you have credentials, or contact Lorikeet support for setup help.

No public MCP server endpoint or open-source client package is published by Lorikeet at this time.

• Let a Lorikeet support agent call an internal MCP server to look up customer account state before answering a billing question
• Connect Lorikeet to a Stripe or Shopify MCP server so the agent can issue refunds or check order status during a chat ticket
• Wire a knowledge base MCP server (Notion, Confluence, Guru) into Lorikeet so the concierge can pull policy text into responses
• Give a Lorikeet voice agent access to an internal operations MCP server to dispatch actions like resetting a password or pausing a subscription
• Use MCP tools to extend Lorikeet workflows for regulated fintech or healthtech use cases that need custom backend calls

• "Resolve this Zendesk ticket: the customer says their last payment failed, look up their Stripe account and retry the charge."
• "When a customer asks about plan limits, pull the latest pricing page from our Notion knowledge base before replying."
• "If a refund is approved by the human reviewer, call the Stripe MCP tool to process it and post a confirmation back in Intercom."
• "Authenticate the caller, check their account status via our internal MCP server, then escalate to a human if KYC is incomplete."
• "Triage incoming Front tickets and tag them with the right product area using the taxonomy MCP tool."

• Purpose-built for complex, action-taking customer support workflows, not just FAQ deflection
• Omnichannel coverage across chat, email, voice, SMS, and WhatsApp with ticketing system integrations (Zendesk, Intercom, Salesforce, Front)
• Treats MCP as a first-class integration method alongside APIs and webhooks, making it easy to plug in custom tools
• Used by fintechs and other regulated industries, so compliance and auditability are part of the platform

• Lorikeet is primarily an MCP client (consumes MCP servers), not a public MCP server you can point Claude Desktop or Cursor at
• Developer documentation is gated behind a login, so setup details are not publicly verifiable
• No open-source repository or published config schema for the MCP integration; setup happens inside the Lorikeet workspace

• Intercom Fin: AI agent for customer support that runs on top of Intercom's inbox
• Zendesk AI Agents (Ultimate): Zendesk's native AI agent product for ticket automation
• Decagon: AI customer support agents with backend action capabilities, similar positioning to Lorikeet