Agents

Create and configure AI voice agents with custom prompts, voices, and behaviors.

Agents

Agents are the core building block of Callem Studio. An agent is an AI-powered voice assistant configured with a system prompt, a voice, a language model, and optional integrations like knowledge bases and tools.

Creating an Agent

1. Navigate to Build > Agents in the sidebar
2. Click New Agent
3. Give your agent a descriptive name

Agent Configuration

Each agent is defined by an Agent Configuration that controls its behavior:

System Prompt

The system prompt is the most important part of your agent. It defines the agent’s personality, role, instructions, and boundaries.

Write your system prompt as if you were briefing a real employee. Be specific about what the agent should do, how it should respond, and what it should avoid.

Example System Prompt

You are Sophie, a friendly receptionist at Acme Dental Clinic.
Your role is to answer incoming calls, schedule appointments, 
and answer questions about our services.

Rules:
- Always greet the caller warmly
- Ask for their name and reason for calling
- If they want an appointment, check availability
- Never provide medical advice
- If unsure, offer to transfer to the office manager

Voice (TTS)

Choose a text-to-speech voice from supported providers:

Provider	Voices	Highlights
ElevenLabs	100+	Natural-sounding, multilingual, custom cloning
Azure	400+	Wide language coverage, low latency

You can configure additional TTS settings like model selection depending on the provider.

Speech-to-Text (STT)

Select the transcription engine for understanding caller speech:

• Deepgram — fast, accurate, supports multiple languages
• Azure Speech — enterprise-grade, wide language support

Language Model (LLM)

Pick the underlying AI model that powers your agent’s reasoning:

• GPT-4o, GPT-4o-mini
• Claude 3.5 Sonnet
• Gemini
• And more, depending on your configuration

Language

Set the primary language for your agent. This affects STT transcription, TTS output, and prompt behavior.

Background Sound

Add ambient background audio to make calls feel more natural (e.g. call center ambiance). You can:

• Enable/disable the background sound
• Choose from a library of pre-configured sounds
• Adjust the volume level

Knowledge Base

Attach one or more knowledge bases to give your agent access to domain-specific information. The agent will automatically retrieve relevant content during conversations.

Tools

Assign tools (custom functions) that your agent can invoke during a call — like booking appointments, checking order status, or transferring the call.

Agent Settings Summary

Setting	Description
Name	Display name for the agent
System Prompt	Instructions that define agent behavior
Voice	TTS voice provider and voice selection
STT	Speech-to-text engine
LLM	Language model powering responses
Language	Primary conversation language
Background Sound	Ambient audio during calls
Knowledge Base	Attached knowledge sources
Tools	Functions the agent can call

Best Practices

Write specific, actionable prompts

Vague prompts like “be helpful” lead to inconsistent behavior. Instead, describe exact scenarios: “When a caller asks about pricing, quote from the attached knowledge base.”

Test with real conversations

After configuring your agent, call the assigned phone number yourself. Listen for unnatural pauses, incorrect responses, or missed instructions.

Start simple, iterate

Begin with a minimal prompt and a single knowledge base. Add tools and complexity once the basic flow works well.