Why use Speak AI with Windsurf

Windsurf is an AI-powered IDE built by Codeium. It goes beyond autocomplete with Cascade, an agentic AI assistant that can reason across your entire codebase, execute multi-step tasks, and integrate with external tools through the Model Context Protocol (MCP). For developers who work with recorded conversations, interview data, or meeting notes, Windsurf's MCP support opens up a direct connection to tools like Speak AI.

MCP (Model Context Protocol) is an open standard that lets AI assistants connect to external data sources and tools. Windsurf's Cascade AI supports MCP natively, which means you can install an MCP server and Cascade immediately has access to that server's capabilities. The Speak AI MCP server provides 81 tools covering transcription, NLP analytics, media management, AI Chat, meeting scheduling, and more. Once installed, Cascade can search your transcripts, run analysis, manage recordings, and export data, all through natural language conversation.

What Speak AI adds to Windsurf

On its own, Windsurf helps you write and understand code. With Speak AI connected, it also helps you understand the conversations that inform your code. You can search across every meeting, interview, and call your team has recorded in Speak AI. Ask Cascade to find what a stakeholder said about a specific feature. Pull sentiment analysis on customer feedback calls. Get keyword and topic breakdowns from sprint retrospectives. The 81 tools in the MCP server cover the full Speak AI platform: transcription in 70+ languages, NLP text analysis, folder and media management, webhook configuration, meeting bot scheduling, and AI Chat queries across your entire library.

Setup takes under five minutes

Install the @speakai/mcp-server package from npm, run the init command, and paste your API key. The init script auto-detects Windsurf and writes the configuration file for you. No manual JSON editing, no restart required. Once connected, open Cascade and start asking questions about your Speak AI data. The server is open source on GitHub under the MIT license.

Practical workflows

The most common use case is searching meeting transcripts while coding. Instead of switching to a browser, finding the recording, and scrubbing through audio, you ask Cascade: "what did the product team decide about the notification system?" and get an answer with timestamps and speaker attribution. For teams that conduct user research interviews, the integration lets developers access research interview transcripts directly in their editor. They can search for pain points, feature requests, and usability issues across dozens of interviews without leaving their IDE.

Beyond search, the MCP server supports uploading files for transcription, scheduling meeting bots to automatically capture upcoming calls, creating and organizing folders, exporting data, and managing webhooks. If your workflow touches audio, video, or conversation data in any way, the Windsurf integration puts it within reach of your normal development environment. Combined with Cascade's ability to reason about your codebase, you can reference real meeting data when writing code, building documentation, or reviewing pull requests.