Bolti AI - Automated Call Agent Platform Logo

How to Deploy Bolti Voice Agents Using the REST API (2026 Guide)

Dhiraj··Updated 6 August 2026

Founder of Bolti, writing about voice AI for Indian businesses.

Bolti, a voice AI platform for phone agents, is built to make conversational voice technology accessible without the traditional engineering overhead. While the Bolti dashboard offers a visual interface to configure your agents, everything you see is backed by a production-ready REST API. Whether you want to spin up agents dynamically, automate outbound calls from your CRM, or sync call logs, the bolti voice ai api gives you complete programmatic control.

With pricing starting at a transparent ‡6/minute pay-as-you-go model (and a free trial with 50 minutes), developers can transition from local testing to automated production systems seamlessly.

How do you deploy Bolti voice agents using the REST API?

To deploy and interact with your voice agents programmatically, you must authenticate your requests using a Personal Access Token (PAT) and interact with Bolti's open API endpoints. Every action available in the Bolti dashboard can be executed via standard HTTP requests. This architecture allows you to integrate voice capabilities directly into your existing software stack, CRM, or custom internal tools.

To begin automating your voice workflows, follow these three core steps:

  1. Generate your credentials: Go to your Bolti dashboard to create an access token.
  2. Configure your agent payload: Define your STT, LLM, and TTS providers via JSON.
  3. Trigger the API call: Send a POST request to create the agent or initiate an outbound call.

Step 1: Generate your Personal Access Token

Before making any API requests, you need an access token to authenticate your server.

  • Log in to your Bolti dashboard.
  • Navigate to your profile at the bottom-left of the sidebar.
  • Click on Access Tokens and select Create Token.
  • Copy the token immediately and store it securely (it will only be shown once).

Note: If you are building locally with AI assistants, you can also use this token to set up the Bolti MCP server in Cursor or Claude Code, allowing you to manage agents directly from your editor.

Step 2: Define your agent configuration

When deploying an agent via the REST API, you must specify the providers that power the voice experience. Bolti allows you to mix and match providers for each of the four core pillars of a voice call:

  • Speech-to-Text (STT): Converts customer speech to text. Choose Deepgram for general English, or Fennec for Indian languages like Hindi, Tamil, and Telugu.
  • Large Language Model (LLM): The brain of your agent. Use fast models like Groq Llama for low-latency scripted flows, or OpenAI GPT-4o for complex reasoning.
  • Text-to-Speech (TTS): Generates the agent's voice. Choose ElevenLabs for ultra-realistic voices, or SarvamAI (Anushka) for high-quality Indian-language output.
  • Telephony: Carries the call. You can buy numbers directly from Bolti or bring your own SIP trunk (Twilio, Plivo, or Exotel).

Here is an example JSON payload to create a new multilingual customer support agent via the API:

{
  "name": "Support Agent",
  "prompt": "You are a helpful customer support agent for a retail brand in India. Speak politely and assist with order tracking.",
  "stt_provider": "fennec",
  "stt_language": "hi",
  "llm_provider": "openai",
  "llm_model": "gpt-4o-mini",
  "tts_provider": "sarvamai",
  "tts_voice_id": "anushka"
}

Step 3: Send the API request to deploy

You can deploy this agent by sending a POST request to the Bolti API. Here is how to do it using curl:

curl -X POST https://api.bolti.co.in/v1/agents \
  -H "Authorization: Bearer YOUR_PERSONAL_ACCESS_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "Support Agent",
    "prompt": "You are a helpful customer support agent. Help the caller track their package.",
    "stt_provider": "deepgram",
    "stt_language": "en",
    "llm_provider": "groq",
    "llm_model": "llama-3-8b",
    "tts_provider": "cartesia",
    "tts_voice_id": "blake"
  }'

A successful response will return a JSON object containing your new agent's unique ID, which you can use to trigger outbound calls or assign to inbound phone numbers.

How do you trigger outbound calls programmatically?

Once your agent is deployed, you can trigger outbound phone calls automatically when an event occurs in your system—such as a user signing up, an abandoned cart, or an appointment reminder trigger.

To place a call, send a POST request to the calls endpoint, specifying the agent ID and the recipient's phone number in E.164 format:

curl -X POST https://api.bolti.co.in/v1/calls \
  -H "Authorization: Bearer YOUR_PERSONAL_ACCESS_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "agent_id": "AGENT_ID_FROM_STEP_3",
    "to_number": "+919876543210",
    "from_number": "+9111XXXXXXXX",
    "custom_variables": {
      "customer_name": "Amit",
      "order_id": "98765"
    }
  }'

Using custom_variables allows you to pass context from your database directly into the call, letting your agent greet the caller by name and reference specific transactional details dynamically.

Why use the Bolti API instead of building from scratch?

Building a custom voice pipeline requires gluing together multiple APIs (STT, LLM, TTS, and Telephony), managing web sockets, and handling packet loss. The bolti voice ai api manages this entire orchestration layer for you.

  • Sub-second latency: Bolti's runtime is optimized for real-time telephony, ensuring turn-taking feels natural and responsive.
  • Interruption handling: If a customer speaks while the agent is talking, Bolti immediately stops the TTS stream to listen, mimicking natural human conversation.
  • Telemetry & Logs: Retrieve transcripts, call duration, and latency metrics programmatically to sync back to your CRM or database.

If you want to test your prompts and settings visually before writing any code, you can easily configure your agent in the web dashboard first, then use the API solely to trigger calls. Alternatively, you can create a free developer account to get API keys and begin testing immediately.

Set up your first voice agent with the API

Deploying automated voice agents doesn't require complex telephony engineering. With Bolti, you can build, configure, and launch production-grade voice workflows in minutes using a single API interface.

Ready to write your first integration? You can create a free developer account and get 50 free calling minutes to test your API integrations. If you need help with custom SIP integrations, enterprise deployment options, or high-volume outbound calling, feel free to contact our team for dedicated developer support.

Frequently Asked Questions

What is the pricing for using the Bolti Voice AI API?

Bolti offers a pay-as-you-go pricing model starting at ‡6 per minute. New developer accounts get 50 free minutes upon signup to test API integrations.

Can I bring my own SIP trunk when using the API?

Yes. Bolti supports Bring Your Own Carrier (BYOC). You can connect your existing Twilio, Plivo, Exotel, or custom SIP trunk directly to your Bolti workspace and trigger calls using the REST API.

Which programming languages are supported by the Bolti API?

Since Bolti exposes a standard REST API, you can make requests using any programming language that supports HTTP requests, including Node.js, Python, Go, Ruby, and PHP.

Does Bolti support multilingual voice agents via the API?

Yes. You can programmatically configure your agents to support Hindi, English, Marathi, Tamil, Telugu, and over 80 other global languages by selecting the appropriate STT and TTS providers in your API payload.