Voice Command Ops

Gmail, Slack, Linear—
still open on the train.

Aurex: Voice Command Ops is an iPhone app that runs your Gmail and Google Calendar by voice. Say what you want — triage the inbox, send a reply, move a meeting — and Aurex does it, then speaks the result back.

Free to explore · No account needed to start

<1s
voice-to-voice
10+
services connected
0
keys on device
Aurex — Your voice runs every app

What Aurex is

An iPhone app that runs your inbox and calendar by voice.

Aurex: Voice Command Ops is a voice assistant for iOS. You hold the orb, say what you want in plain language, and Aurex carries it out in the accounts you have connected — then speaks the result back. There is nothing to type and no dashboard to learn.

Connecting a Google account is optional, and it is what makes the two things below possible. Aurex only ever acts on an instruction you gave it; it does not read your mailbox in the background for any other purpose.

With Gmail

Aurex reads your mail so it can triage the inbox out loud — who wrote, what they want, and what is waiting on you. It drafts, sends, replies to, labels, and archives messages you dictate. It cannot permanently delete mail.

With Google Calendar

Aurex reads your calendars and events to answer questions about your schedule and find open time, and it creates or updates events when you ask for one out loud.

Slack, GitHub, and Notion connect the same way, each optional and each disconnected independently. What Aurex does with any of this data — and how to delete it — is spelled out in the privacy policy.

Philosophy

Speed is not a feature. It's the product.

Every existing assistant makes you wait — for transcription, for synthesis, for a route that should have been obvious. Aurex uses native speech-to-speech at the model layer, attaches your MCP connections directly to the voice session, and narrates while it works. The silence between speaking and hearing an answer is the enemy. We obsess over that gap.

What it replaces

You open Slack, then Gmail, then Linear. Thirty seconds later you're still reading thread context instead of making the decision. Aurex speaks the summary before your screen is even lit.

Aurex — Inbox, Slack, tasks in one spoken ask

How it works

Three seconds from pocket to answer.

01

Speak the command

Push-to-talk or wake word. "What shipped in the KombuVault repo yesterday?" Barge-in at any point — Aurex stops talking the moment you start.

02

Aurex reasons and routes

The voice session selects the right MCP tool — GitHub, Linear, Slack, whatever fits — and narrates while it calls. No silence. No spinner.

03

Hear the answer

Voice-to-voice in under a second on repeat turns. For write commands, Aurex reads the action back and waits for your spoken confirmation before executing.

Built for the operator

Adapts to how you already run.

Onboarding shapes your command priority

Aurex asks once what you want from it — faster app-switching, deep voice ops, sharper daily workflow. Your answer determines how it weights tool selection and narration depth.

Read and write, same voice

"What's on my calendar this afternoon?" and "Reply to Sarah, I'll call at 3" use the same interface. Reads are instant; writes require a spoken yes for anything consequential.

Barge-in, always

You can interrupt mid-sentence. Aurex stops immediately and pivots. The assistant never holds the floor.

Aurex onboarding — tailored to how you work

Integrations

Every tool you already use.

Aurex attaches MCP servers directly to the voice session. No rebuild of your existing integrations. Your credentials stay server-side.

Gmail
Slack
Linear
Asana
Notion
Calendar
Drive
Mercury
Typefully
Kiloforge MCP
Aurex — Connect every tool you already use

Your API keys never touch the device.

The only backend component is a token broker that mints short-lived session tokens. MCP credentials live server-side, co-located with the OpenAI Realtime endpoint for minimum latency. Your phone receives a credential that expires with the session.

~350ms
p50 voice-to-voice (repeat turns)
~1.2s
first turn glass-to-glass
~25%
p95 latency reduction in gpt-realtime-2.1-mini
0
keys exposed on device

FAQ

The questions that matter.

Aurex app icon

Get notified when Aurex launches.

We're in a closed beta. Leave your email and we'll reach out when a spot opens — no spam, no newsletter.

Closed beta · iOS · Free to explore

Command Line

Thinking out loud about voice, speed, and the future of ops.

Ultimate Guide

How to Control Gmail, Slack, Linear, and Notion Entirely by Voice in 2026

Most productivity apps still assume your hands are free and your screen is in front of you — but a new generation of voice-native tools is changing that. This guide walks through exactly how a single operator can wire up every major SaaS tool and fire off read and write commands hands-free, from anywhere. From push-to-talk to confirmation gates on high-stakes actions, here's what a real voice ops setup looks like in 2026.

Read more →9 min read
Engineering

The Hidden Latency Killers in AI Voice Assistants (And How to Engineer Around Them)

Sub-second voice response isn't just about picking a fast model — it's about eliminating every silent millisecond tax hiding in your stack. From misconfigured VAD thresholds that cut off speech too early, to slow MCP tool schemas that stall reasoning, the real latency enemies are often invisible until you measure them. This post names each one and explains the engineering fix.

Read more →9 min read