Yokwi · a voice for the agent you already run

It hears every word.
It waits for you.

Yokwi lives on your iPhone and talks to your AI agent, the one on your Mac, your PC, or a server you rent. It listens, it carries your words as text to your machine, and it reads your agent's spoken explanation back in a Kokoro voice you chose, on the phone itself. Your hands stay free.

One question, start to finish · a scripted illustration

Watch how it works.

You ask. Your agent goes off and does the work. When it reports its steps, Yokwi shows each one while you wait. Then Yokwi reads the spoken explanation your agent sends with its answer.

Ready
your agent0.0s

Prewritten illustration. In the app, the steps shown depend on what your agent reports, and the answer is your agent's own.

What it actually does

All this little thing does
is talk to your machine.

There's no Yokwi brain in the cloud. Your phone hears you, it goes through our server, and the agent running on your computer does the thinking and the talking. Press a phrase and watch it go round.

Your iPhoneHears you. Apple's Speech framework turns speech into text, on the device or on Apple's servers depending on device and language. Reads the reply with its own on-device Kokoro voice.

usernamethrough our server

Your unique usernameYokwi sits on it and carries your words, as text, to your machine, then brings the answer back. It routes what it carries and keeps only what delivery needs, briefly.

$ agent

Your machineYour agent, your model, your API key. Hermes, OpenClaw, or your own, any agent you already run. Sometimes it takes a moment.

Example replies. Your agent gives its own.

On your phone

Microphone, Apple speech recognition into text, your Yokwi, and its on-device Kokoro voice. The app never needs an account or a password.

Your unique username

One username per person, kept while your address and credentials stay valid. Text and control messages pass through our server, a wire between your phone and your machine, and nothing more.

On your machine

Everything that thinks. Which model, which key, what it's allowed to do, all yours, all where you can see it.

Six screens, one hello

Setup is the whole onboarding.
Here it is.

Illustrations of the onboarding screens. You choose a username, generate your Yokwi, send the generated pairing instructions to your agent (two short messages, in order), and run one voice test. Pairing is then kept across ordinary reconnects while your address and credentials stay valid.

1 · Welcome

Yokwi

It hears every word.
It waits for you.

Say hello
2 · Permissions

First, let Yokwi hear you

Your voice is transcribed by Apple's Speech framework (Apple may process recognition for some devices and languages). Yokwi never sends your raw audio to its relay or to your agent.

Microphone
Speech Recognition
Allow Microphone & Speech
Yokwi never sends raw audio to its relay or to your agent.
3 · Your username

Choose your username

Create your unique username, your agent reaches you here while your address stays claimed.

username
Claim My Address
4 · Design your Yokwi

Design your Yokwi

Every Yokwi grows its own aura, orbit, and quiet little markings.

Generate Your Yokwi
5 · Pair your agent

Meet your Yokwi

Send the generated pairing instructions to your AI agent. Once it runs the pair command, it connects to your Yokwi address.

App connected
Agent connected
Pairing confirmed
Start Voice Test
6 · Voice test (shown after a test actually completes)

Voice test passed

The app and agent are paired. Your voice channel is ready.

Pairing is retained while your address and credentials stay valid.
Start Talking
Step four, for real

Design your Yokwi

Five siblings, Yokwi, Dexi, Luxi, Mari and Luma, each with its own aura, orbit, gaze and quiet little markings. Roll one. This is the same generator the app uses, same odds and all.

Every Yokwi grows its own aura, orbit, and quiet little markings.

Until you roll, it keeps morphing, a little teaser of what the generator can make.

Your voice, your machine

Yokwi holds the microphone
so your hands stay free.

Nothing here is trying to learn you. The thinking happens on hardware you control, with a model and a key you picked.

Transcribed by AppleApple's Speech framework turns your voice into text; depending on your device and language, Apple may process recognition on its servers under Apple's policy. Only that text goes to your agent.
Sent only to your agentEach finished turn goes through our server to your machine as text, never audio. Nowhere else.
No account. No password.You claim a name. That's the whole sign-up. There's no email to leak and no profile to build.
Your model, your rulesA hosted model or one on your own machine, whatever your agent runs on, its data policy is the one that applies. You chose it.

Read the full privacy policy →

Plans

Try it free. Keep it with Plus.

Download and set up for free. Your first conversation starts seven days of Plus access; after that, you choose whether to subscribe. Drive, the car version, is in testing with early drivers.

Free

$0

  • Free download and setup
  • Seven days of Plus access when you start using conversations
  • Your Yokwi and its family

Plus 7 days free first

Monthly price shown in the App Store

  • After your free week, choose whether to subscribe
  • Every Yokwi voice
  • Keep talking with your agent

Drive Trial and testing

In testing

Nothing is bought when the free week starts or ends. Subscriptions are billed by the App Store and can be cancelled there any time. Pro, for advanced agents, comes later.

Say hello.

Free to download. Bring the agent you already have. Pairing takes the generated instructions once and is kept across ordinary reconnects while your address and credentials stay valid.