Configure Your Voice Agent

Each agent combines four components: a system prompt that defines behavior, an LLM for reasoning, a voice for speech, and optional tools for actions. Here’s how they fit together.

Agent components

Component	Required	What it does
System prompt	Yes	Defines personality, role, rules, knowledge
LLM config	Yes	Which AI model powers the brain (OpenAI, Groq, etc.)
Voice config	Yes	Which voice speaks the responses (ElevenLabs, Cartesia, etc.)
Tools	No	External actions the agent can take mid-conversation
Schedule	No	Business hours — when the agent accepts calls
Webhooks	No	URLs to notify when things happen

Agent states

An agent is either enabled or disabled. Enabled — Active and handling conversations. This is the default. Configuration changes apply immediately, even to active calls. Disabled — Paused. Won’t accept new conversations. Existing calls complete normally. Configuration is preserved. No charges while disabled.

API example: Disable an agent

await fetch(`https://blackbox.dasha.ai/api/v1/agents/${agentId}`, {
  method: 'PUT',
  headers: { 'Authorization': 'Bearer YOUR_API_KEY', 'Content-Type': 'application/json' },
  body: JSON.stringify({ isEnabled: false })
});

Changes to enabled agents apply immediately. If you’re making significant changes, disable the agent first, make changes, test, then re-enable.

What’s next

Conversations

Learn about conversation types and states

Tools

Let your agent take actions

Create your first agent

Step-by-step guide

LLM Configuration

Model selection and parameters

Overview Conversations

⌘I

Introduction

Create

Test

Deploy

Monitor

Advanced Features

Troubleshooting

Agent components

Agent states

What’s next

Conversations

Tools

Create your first agent

LLM Configuration

​Agent components

​Agent states

​What’s next

Conversations

Tools

Create your first agent

LLM Configuration

Agent components

Agent states

What’s next