Build on DialPress

The layer under products that put AI on the phone

If your product answers or places calls with AI, DialPress is the layer underneath: numbers and carriers, the media path, speech, and an agent runtime you can replace piece by piece.

Talk to an engineer

Read the developer overview

An AI voice botbot.js

const dialpress = require('dialpress') const bot = dialpress() bot.invite('/ai/:name', async (req, res) => { await res.answer() const media = await res.media({ sampleRate: 16000 }) media.on('audio', (pcm) => stt.write(pcm)) // caller voice llm.on('reply', async (t) => media.play(await tts(t))) vad.on('speech', () => media.clear()) // barge-in res.on('dtmf', (d) => d === '0' && res.handoff('/internal/0', { summary: llm.summary() })) }) bot.connect({ address: 'agi.dialpress.ai', pathPrefixes: ['/ai'] })const dialpress = require('dialpress') const bot = dialpress() bot.invite('/ai/:name', async (req, res) => { await res.answer() const media = await res.media({ sampleRate: 16000 }) media.on('audio', (pcm) => stt.write(pcm)) // caller voice llm.on('reply', async (t) => media.play(await tts(t))) vad.on('speech', () => media.clear()) // barge-in res.on('dtmf', (d) => d === '0' && res.handoff('/internal/0', { summary: llm.summary() })) }) bot.connect({ address: 'agi.dialpress.ai', pathPrefixes: ['/ai'] })

Summary

Direct

Audio goes from the carrier to your model without a third-party hop.

Replaceable

Use our speech and agents, or plug in your own models.

Compliant

STIR/SHAKEN, 10DLC, and E911 handled at the edge.

Portable

Our cloud, or self-hosted for customers who need it.

A model is not a phone system

A voice AI product is judged on the call: how fast the agent answers, whether the number is trusted, whether a transfer keeps the caller. Those are network problems. Building on DialPress means your team works on the product, not on carriers and media servers.

Why voice AI needs its own stack

How it works

Numbers and routing

Give each of your customers numbers, routed to the right agent path.

Your pipeline

Stream call audio to your own recognition, LLM, and text to speech, or use ours.

Handoffs

Pass live calls to people with res.handoff(), without a transfer.

Operations

Events, CDRs, SIP traces, and an MCP server for your own tooling.

Built from

The DialPress primitives this uses.

01

Voice Gateway

SIP trunks, phone numbers, and WebRTC in one gateway, with STIR/SHAKEN, 10DLC, and E911 at the edge.

02

Media Streams

The live audio of any call, in your code, both ways, in real time.

03

Speech

Streaming recognition, text to speech, and keyword spotting, on your GPUs or a cloud provider.

04

Agents

Voice and chat agents that answer from your knowledge base and call your APIs over MCP.

05

Messaging

SMS, email, SIP MESSAGE, and HTTP behind one message model.

06

Softphone

A client for the web, iOS, and Android on your own SIP accounts.

Common questions

Can I use my own LLM and speech models?

Yes. Stream audio to your own pipeline, or point DialPress at the models you choose.

Can you run in my customers’ clouds?

Yes. DialPress self-hosts on Linux, Docker Compose, or Kubernetes.

How is it priced for platforms?

Volume and self-hosted deployments are quoted. Talk to us about your traffic.

Build your product on the voice layer.

Tell us about your traffic and where it has to run.

Talk to an engineer

Read the developer overview