Agents Advanced

D-ID Agents combine streaming avatars with LLM-powered conversation to create interactive AI characters that users can talk to face-to-face in a web browser. Agents handle the complete pipeline: listening, thinking, and responding with a visible, speaking face.

Agent Architecture

A D-ID Agent connects four components in a real-time loop:

Creating an Agent

1. Choose a face

   Select a source image for the agent's appearance. This determines the visual identity.

2. Configure the LLM

   Connect your LLM provider and write the system prompt that defines the agent's personality, knowledge, and behavior.

3. Add a knowledge base

   Upload documents, FAQs, or URLs that the agent can reference during conversations for accurate, grounded responses.

4. Set voice and language

   Choose the TTS voice that matches the agent's persona. Configure supported languages.

5. Deploy

   Embed the agent on your website with an iframe or integrate via the Agents API.

Use Cases

• Customer Support: 24/7 AI representative that answers questions with a friendly face
• Education: Interactive tutor that explains concepts and answers student questions
• Sales: Product advisor that recommends items based on conversation
• Healthcare: Patient intake assistant that collects information conversationally