AI voice agent infrastructure for developers.
Orbitali gives developers a realtime voice-agent runtime with prompts, tools, transcripts, knowledge, and usage tracking in one platform. You own the business logic. We run the agent layer.
Can I move my appointment to Friday?
I can help with that. Let me check availability.
{
"date": "2026-05-15",
"service": "consultation",
"caller_id": "+1555..."
}Talk to an Orbitali agent directly from the browser.
The page asks this Next.js backend for a short-lived realtime session, then streams microphone audio to the agent over WebSocket and plays the returned voice audio.
Your browser will request microphone access. The long-lived Orbitali API key never leaves the server.
A programmable AI voice agent runtime, not another chatbot builder.
Define the agent, connect a voice channel, configure one Server URL, and route conversations through a low-latency realtime agent stack. Orbitali stays generic so your application logic can stay in your own API.
Voice channels
Connect carrier channels to agents without making telephony operations the center of your product.
Realtime voice stack
Speech recognition, reasoning, and voice output run through one low-latency conversation loop.
Assistant requests
Fetch caller-specific context from your backend before Orbitali answers the call.
Webhook tools
Expose booking, lookup, support, payment, or CRM actions as named tool-call events.
Knowledge base
Upload docs for FAQ, policy, and product answers without building your own retrieval layer.
Logs and usage
Inspect transcripts, tool invocations, token usage, duration, and call outcomes.
From live conversation to tool execution in one path.
Every conversation is routed to an agent, resolved into a prompt, streamed through the voice runtime, and recorded with transcript and tool activity.
Inbound voice
A carrier channel sends the conversation to Orbitali's realtime agent service.
Route agent
The channel maps to your configured agent and prompt settings.
Resolve context
Orbitali uses static instructions or sends assistant-request to your Server URL.
Stream conversation
The voice runtime speaks, listens, reasons, and handles interruptions.
Call tools
Your Server URL receives tool-call events, executes business logic, and returns results.
Record everything
Transcripts, tool calls, usage, and status are saved for review.
Prompts and tools flow through one webhook contract.
Use static identity and instructions for simple agents, or let Orbitali send an assistant-request event to your Server URL before the call is answered. During the call, custom tools are delivered to the same Server URL as tool-call events with typed arguments and predictable timeouts.
A tool is a webhook event contract
Describe when the agent should call it, define the arguments, and Orbitali posts tool-call events to the agent's Server URL during the conversation.
{
"name": "check_availability",
"description": "Check available appointment slots.",
"parameters": {
"type": "object",
"properties": {
"date": { "type": "string" },
"service_id": { "type": "string" }
},
"required": ["date", "service_id"]
},
"timeout_ms": 10000
}assistant-request
Use caller metadata to assemble the exact prompt and greeting your agent needs before the call is answered.
POST https://api.acme.com/orbitali
{
"message": {
"type": "assistant-request",
"call": {
"id": "uuid",
"fromNumber": "+15559876543",
"toNumber": "+15551234567"
}
}
}Predictable pricing for deployed voice agents.
Connect your own Telnyx setup or use an Orbitali-managed carrier configuration. Every plan is billed in 1-minute intervals for handled inbound voice conversations.
BYO Telnyx
For teams that already manage their own Telnyx account and voice channels.
Managed number
For businesses that want Orbitali to handle the carrier configuration.
Orbitali is focused on inbound calls today. Outbound campaigns and regulated calling workflows are not currently offered. Prices are plus applicable tax.
Building voice agents for real customer calls?
Book a call with us to discuss what you want to deploy. We are prioritizing developers with concrete AI voice agent use cases.