Connect · Automate · Build

Wave for Agents

Bring the conversations you capture in Wave into the AI workstreams where your next decisions and actions happen.

Choose your path

Your meeting memory should work everywhere.

Start with the way you want to work: connect an assistant, automate locally, or build something custom. One Wave account, three purpose-built paths.

Connect · MCP

Connect your AI assistant

Best for: Claude, ChatGPT, and other MCP clients

Bring your Wave meeting memory into Claude, ChatGPT, Claude Code, and other MCP clients. One URL, then sign in with Wave.

Search sessions and retrieve exact, speaker-labeled transcripts
Organize folders and safely keep action items in sync
OAuth-scoped access you can review and revoke

Connect Wave Read the MCP guide

Automate · CLI

Automate from the terminal

Best for: Local agents, scripts, and scheduled jobs

Search, organize, export, and react to Wave sessions from local agents, scripts, cron jobs, and shell workflows.

npm install -g @waveai/cli
Human-readable output for you; JSON output for your tools
Events, exports, folders, transcripts, and action-item workflows

Read the CLI guide View on npm

Build · API

Build a custom workflow

Best for: Products, backends, and custom integrations

Use the Wave API when your AI workstream needs custom logic, a backend integration, webhooks, or complete control over the experience.

Sessions, transcripts, summaries, folders, action items, and media
Event feeds and signed webhooks for newly completed recordings
Scoped bearer tokens and a complete OpenAPI reference

Read the API guide Create an API token

What’s possible

AI workstreams, with your real context

Generic AI is useful. AI that understands the meetings you’ve actually had can prepare, recall, organize, and follow through with you.

Ask your AI anything about anything you've said

Give Claude, ChatGPT, or any MCP-capable agent live access to your transcripts and summaries. It stops needing you to paste context.

Automate the post-meeting work

An agent listens for every new recording, pulls the action items, updates your CRM, drafts the follow-up, and asks you to approve. You stop doing busywork.

Search every conversation you've ever had

Semantic search over years of recordings. Ask in natural language, get ranked snippets and the sessions they came from.

Pipe raw data anywhere

Full transcripts with speaker attribution, structured summaries, action items, signed audio URLs — as JSON, as text, as whatever your downstream tool needs.

The data is yours. Always.

Every transcript, every speaker-labeled segment, every summary, every action item, every byte of audio — exportable, queryable, and automatable. Use it with whatever AI you trust today or whatever comes next. Wave keeps your meeting memory portable, with you in control of where it goes.

TranscriptsSpeaker-labeled textAI summariesAction itemsAudio & videoSemantic searchFoldersWebhooks

Works where you work

One meeting memory. Familiar tools.

ClaudeMCPWeb · desktop · mobile

ChatGPTMCPApps · connectors

Claude CodeMCPTerminal

CursorMCP

WindsurfMCP

ZedMCP

Local agentsCLIHuman + JSON output

ZapierAPIOAuth

Your own codeAPIHTTPS + bearer token

Connect through MCP, automate with the Wave CLI, or build over HTTPS. The label on each tool shows the path it uses.

How it works

From connection to useful context in three steps

Step 1

Choose a path

Connect with MCP, automate with the CLI, or build with the API.

Step 2

Authorize Wave

Step 3

Put context to work

Ask, automate, or build. The authoritative guides live at api.wave.co.

Your context, your control

Your meetings should work for you.

Connect Wave to the AI tools you already trust, automate the work around them, or build a workstream of your own.

Connect Wave Read the technical guides