Agents
What you'll learn
- The agent types in AutoTalk: LLM agents and Typebot (flow-based) agents
- When to use each type
Agents are automated assistants that handle conversations on your behalf. The Assistant type field offers three options: llm, typebot, and actor. This page covers the first two:
- LLM Agents use large language models to generate dynamic, context-aware responses. They can understand context, call tools, and adapt to different situations.
- Typebot agents follow pre-designed, flow-based conversation scripts with buttons, menus, and branching logic.
The third option, actor, runs ordered rule-based steps with CEL conditions on each message.
LLM Agents vs. Typebot Agents
| Feature | LLM Agents | Typebot Agents |
|---|---|---|
| Response style | Dynamic, generated | Pre-scripted |
| Setup complexity | Moderate | Visual drag-and-drop |
| Flexibility | High — adapts to any question | Fixed — follows defined paths |
| Best for | Support, sales, complex queries | Surveys, menus, simple flows |
| Tools & actions | Yes — can call APIs, query data | Limited |
Typebot Agents
Typebot is an open-source visual chatbot builder. You design conversation flows using a drag-and-drop editor with blocks for messages, questions, conditions, and integrations.
Setting up a Typebot agent
- Build your chatbot flow in the Typebot editor.
- In AutoTalk, go to Automations > Agents and click the + (add) button.
- Enter a Name for the agent.
- Select typebot from the Assistant type dropdown.
- Fill in the API URL (the full Typebot startChat URL for your flow) and, if your Typebot is private, the optional API token.
- Save the agent. Logging verbosity is set later via Log level under the Advanced tab's Options.
When to use Typebot
Typebot is a good choice when:
- You want a visual, drag-and-drop flow builder
- Your conversation follows a fixed structure (intake forms, surveys, booking wizards)
- You want an open-source solution you can self-host
- Your team prefers designing flows visually rather than writing prompts
tip
You can combine Typebot agents with LLM agents. For example, use a Typebot agent for structured onboarding on your website and an LLM agent for open-ended WhatsApp support.
Next steps
- AI Agents overview — Learn about LLM-powered agents
- Creating an agent — Set up your first AI agent