Joey

Joey uses paid models on the Replicate platform and points out that more capable models can help achieve better content quality.

Any customer who contacts AssemblyAI—whether through the website chat window, by contacting sales, or by emailing the support inbox—is first received by Joey rather than a human.

Joey uses Voyage embeddings to prioritize retrieval of the most relevant documents in order to reduce search time.

Joey raised the end-to-end resolution rate from 10% to 80% in the first week after deployment.

Joey is AssemblyAI's self-developed AI customer service agent, built on the Claude Agent SDK, capable of managing infrastructure, owning a file system, writing and debugging code, and calling tools.

Users cannot directly reach a human unless they proactively ask Joey to escalate; Joey handles 100% of inbound tickets and transfers approximately 20% of cases that require human assistance.

The team added voice capability to Joey using AssemblyAI's own Voice Agent API, chaining speech-to-text, LLM, and text-to-speech together via a single WebSocket connection.

Even when the entire documentation site goes down, Joey is still able to provide accurate answers about the latest features based on local Markdown files.

Joey's total cost (tokens plus inference/infrastructure) is only approximately $700 per month.

AssemblyAI built AI customer service agent Joey using the Claude Agent SDK, raising the end-to-end ticket resolution rate from approximately 10% to 80%.

Joey is the first line of defense for all inbound conversations, covering the website chat window, sales contacts, and the support inbox.

Joey converts documentation, changelogs, pricing pages, and other content into locally stored Markdown files, enabling it to provide up-to-date answers based on local files even when the documentation site is down.

Joey uses Voyage's embedding model for document retrieval because Voyage has targeted optimizations for retrieval precision on technical documentation, code, and API reference content.

Joey is deployed on Railway and can go live with a new version within 30 seconds.

Joey's instructions are provided in the form of a CLAUDE.md file, containing approximately 30,000 lines of guardrails and customer communication guidelines.

In the first week after Joey's deployment, the end-to-end resolution rate jumped from 10% to 80%, with a total cost of approximately $700 per month (tokens plus infrastructure).

Joey handles 100% of inbound tickets and escalates only 20% of them to humans for legitimate reasons.

Joey supports voice mode, chaining speech-to-text, LLM, and text-to-speech over a single WebSocket connection while handling pauses, interruptions, barge-ins, and other interactions.

Creator Joey demonstrated an n8n-based AI Agent automation solution that completes the entire process from content generation to multi-platform publishing by simply triggering instructions in a Google Sheet.

Joey said that if you need to deploy this system in your own cloud account, you can contact him for support.