Create several AI agents — each with its own personality, model, and voice — set a topic, and watch them talk to each other autonomously, with text-to-speech and animated avatars. Or step in as the moderator.
AI Talks lets you create AI characters and watch them hold an autonomous conversation — with realistic voices and animated avatars. Or jump in and steer the discussion yourself.
Add your own API keys in Settings. They are stored only in your browser and sent straight to the provider — never saved on our server.
Tip: give each agent a different model — GPT vs Claude vs Gemini — and a distinct voice.
Stored only in your browser. Sent straight to the provider, never to our server.
This is a playground where several language models play characters and talk to each other. You give each agent a name, a personality (which becomes its system prompt), a model, and a voice — then enter a topic and start the conversation. Agents reply in turn, are spoken aloud with realistic voices, and their emoji avatars lip-sync to the audio and shift expression with the mood of the discussion.
To let the agents think and speak, add your own provider API keys in Settings (OpenAI, Anthropic, Google, and for voice — OpenAI TTS or ElevenLabs). Keys are stored only in your browser and sent straight to the provider — we never store them. That way you control which models and voices run and what it costs.
Why use it?
It is both entertainment and an experiment: it is fun to pit different models (GPT vs Claude vs Gemini) in the roles of different characters and watch a lively discussion emerge — with personalities, emotions, and voices.
Where does my data go?
All the logic runs in your browser; only the reply and voice requests go to the server and are immediately forwarded to your chosen provider with your key. We store neither keys nor conversations.
This is one more tool in the section — all utilities are gathered on the online tools page.
Let’s discuss your project?
Free consultation + cost calculation
