Connect ElevenLabs to your AI agent
Realistic text-to-speech voices.
What it changes
The platform ships with its own models. Connecting your own means a specific voice, a specific image model, or a video provider — on your account and your terms.
What people ask once it is connected
How to connect it
- 1
Paste your api key
Create a key in the provider’s dashboard and paste it in. It is sealed on our side and sent only to that provider.
Stored as
ELEVENLABS_API_KEY - 2
The agent is briefed automatically
Connecting it hands the agent a short description of what ElevenLabs provides and how to call it. You write no routing rules and no prompts.
Voice. POST /text-to-speech/{voiceId} with header xi-api-key. - 3
Ask it something that needs live data
The agent decides a request is needed, makes it, and answers with what came back. If the call fails it says so rather than filling the gap with a guess.
The technical details
- Base URL
- https://api.elevenlabs.io/v1
- Authentication
- API key
- Secret name
- ELEVENLABS_API_KEY
- Category
- AI & Media
Your key or ours. On your own key, calls run against your account and spend no credits. Where we offer a managed key for a connector you can use ours instead and pay per request in credits — priced per connector, refunded when a request fails.
Questions
What can my AI agent do with ElevenLabs?
Realistic text-to-speech voices. Extra models for text, speech, images and video. The agent calls it during a conversation, when the answer needs live data rather than a stored one.
Do I need my own ElevenLabs key?
Either. Paste your own ELEVENLABS_API_KEY and calls run on your account with no credits spent. Where we offer a managed key for a connector, you can use ours instead and pay per request in credits.
Where is my key stored?
Sealed, write-only. It is encrypted before it is written, never returned to any browser — including yours — and only ever sent to the provider it belongs to. You can replace or remove it at any time.
How does the agent know when to use it?
Connecting it adds a short brief to the agent describing what ElevenLabs provides and how to request it. The agent reads that and calls it when a question needs it — you do not write any routing rules.
What if ElevenLabs is down or the key is wrong?
The agent says so and carries on with what it does know, rather than inventing an answer. A failed managed request is refunded.
Give your agent live data
Describe what you do in a sentence, then connect ElevenLabs in the agent’s Connectors tab. It starts using it in the next message.
Build your agent free