The ElevenLabs MCP connector no longer just manages agents, it now makes content as well: speech, transcription, dubbing, music, sound effects, images and video. One install, sign in with your account, no server and no API key.
- The connector works in Claude, ChatGPT, Cursor, Grokbot, Hermes and others and draws on over 50 models.
- Everything generated lands in the ElevenCreative workspace and can be finished in Studio.
- The admin chooses which tools the connection can call, and data residency is picked at connection time.
Anyone who has made an ad from scratch knows how many tabs are open by the end. Text in one, voice in another, music in a third, editing in a fourth, and files moving between them with names like final_final_2.
ElevenLabs is chasing exactly that mess. Instead of you going to the tools, the tools come into the conversation you are already sitting in.
The good part here is less about the generating and more about where it goes. Every clip is saved in your workspace instead of sinking into the chat history, and tomorrow you can find it, open it in the editor and fix it.
The second important part is boring: the admin decides what the connector may do. In a team where one absent-minded question could launch a dub of the entire archive on the company account, that line matters more than the list of models.
One thing is not in the announcement: the price. Generation runs through your account, and an assistant can fire off five versions of the music where you would have made one. Watch the spend in the first week.