Why voice AI needs its own stack


A voice agent is only as good as the call it rides on. Why we built the SIP server, media path, speech, and agent runtime as one stack.
Most AI voice products are a model glued to someone else’s phone line. When the call quality drops, the latency spikes, or the number gets flagged as spam, the model can’t fix it.
Latency is a network problem first
Turn-taking feels human at a few hundred milliseconds. Every hop between carrier, media server, and model eats into that budget, so we keep the media path and the agent on the same infrastructure.
Trust is a carrier problem too
STIR/SHAKEN attestation, 10DLC registration, and E911 aren’t add-ons — they decide whether your calls connect and your texts deliver. They are bundled into every DialPress plan.
And it’s why the price is lower
Owning the stack means we don’t pay a reseller margin on every minute — so neither do you. And because every part is a module, you can run the same stack in your own data center.
Field notes
Keep reading
Build an AI voice bot like an Express app
The dialpress Node SDK turns SIP into routes and handlers. Here is a PBX and an AI voice bot that share one gateway — and pass calls between them.

SIP vs WebRTC for AI voice agents
SIP reaches every phone number on earth; WebRTC reaches every browser. A production voice agent needs both — here is how they fit together.

Giving voice agents tools with MCP
Your booking system, your CRM, your order lookup — any API can become a voice agent’s tool without writing a line of agent code.
