You have probably already met one of these tools. They show up uninvited at the start of a Zoom or Google Meet call, record everything, and email you a summary afterwards. That is fine as far as it goes — but it stops short. A transcript that ends up in your inbox as a summary is a dead end. A transcript that lands inside Claude becomes raw material for action items, dashboards, follow-up emails, cross-meeting analysis, and anything else you can think to do with it. This chapter walks through the landscape, explains the two integration types you will encounter (native Claude connectors and MCP servers), and shows you the two paths for getting a transcript into Claude. The simplest path — paste — works for every tool on the market. The richer path — native connector — works for the tools that have invested in tight Claude integration.

There are many AI transcription tools available, including Otter.ai, Fireflies.ai, Fathom, Zoom AI Companion, Read.ai, and Granola. Most can record meetings, generate transcripts, and create summaries. In Claude, these tools are available either as native connectors or through MCP (Model Context Protocol). Native connectors simply provide a more convenient setup experience, while MCP integrations offer the same core capabilities. In this course, we will use both approaches.

You can work with transcripts in two ways. The simplest is to copy and paste a transcript into Claude and ask for a summary or action items. For a more seamless workflow, connect Otter by asking Claude to guide you through the setup process, then verify the connection by retrieving a recent transcript. Once connected, spend a few minutes brainstorming automation ideas with Claude, such as generating meeting summaries, extracting tasks, or combining insights from multiple conversations.