Building an AI voice agent from scratch is tough. You need speech-to-text, a large language model, text-to-speech, and telephony — plus glue code to connect it all.
But the good news is: today, you don’t need to reinvent the wheel. With platforms like Rounded, you can build, deploy, and scale AI voice agents in minutes, while still keeping full API access for developers.
In this tutorial, we’ll walk through how to:
- Define a simple voice agent.
- Deploy it in a no-code builder.
- Trigger calls via API.
- Pass variables to personalize conversations.
1. Define your agent
First, decide what your agent should do:
- Medical secretary → schedule appointments.
- Lead recall → call back new leads within 1 minute.
- Customer support → answer FAQs and create tickets.
On Rounded, you define the agent’s role, style, and rules directly in a no-code editor.
2. Deploy in minutes
Once the agent is ready, you can:
- Assign it a phone number.
- Configure call duration.
- Select a voice and language (multilingual supported).
No SIP headaches, no carrier setup — Rounded handles telephony for you.
3. Start a call with the API
Now let’s trigger a call programmatically. Rounded exposes a REST API to connect your agent to any workflow.
Here’s a simple example to start a call:
bash
curl -X POST "https://api.callrounded.com/v1/calls" \
-H "Content-Type: application/json" \
-H "Authorization: Bearer YOUR_API_KEY" \
-d '{
"to": "+14155552671",
"agent_id": "agent_12345"
}'
Top comments (0)