Use large models
on your phone.
Running on your PC,
a world away.
Reach it from the next room or a different country. Olace connects every device you own to the AI on your desktop. One encrypted network, your conversations in sync everywhere, under your control. No DIY hacks. No network setup.
[ demo video ]
the split-location shot: a 70B model answering on a phone at a café,
hard-cut to the home GPU. the only color on the page.
Your best model is stuck
on one machine.
The most capable AI you own, running entirely under your control. Leave it behind and it's a locked room you can't get into.
Going local costs you
convenience.
The setup, the upkeep, and conversations that don't follow you from device to device. Real power, with none of the ease.
You could set it up yourself. The quick way, opening a port to your machine, leaves it open to the internet. The safe way, a tunnel or a VPN, is a system you build and keep maintaining. Even then, all you have is a connection to the machine. Not a native way to use your model on every device, with your conversations in sync.
Three steps. Then it just works.
No terminal, no config. Olace can pick and set up the best model for your machine, pair it with your devices, and manage all the complexity underneath.
On your devices
Download the native Olace app on your desktop and on the phone, laptop, or any other device you carry.
Once, and never again
Pair your phone, or any device, with your desktop from the app. They stay paired through an end-to-end encrypted connection, on your LAN or away.
From anywhere
Use your powerful home model from the next room, a café... in a different country. Responses stream back live to your device.
One thread, across
everything you carry.
Close the laptop. The conversation is on your phone before you reach the door. Encrypted and readable only by your devices.
[ continuity clip ]
laptop closes · phone already has the thread
Every device you own.
Not just the matching ones.
Android and iPhone. Windows, Linux, macOS. The web. Ollama and LM Studio. Olace joins it all into one network. All native apps.
[ device wall ]
the same conversation, live on six screens
Reach even higher.
Your desktop can run serious models. When you want a frontier open model on top, Olace serves it from a zero‑data‑retention cloud: nothing you send is trained on or stored. Or bring your own key, and it runs directly through your device.
Free where it should be free.
Local and paired AI, across every device, is free and unmetered. You pay only for continuity and the cloud.
Questions, answered.
How does Olace work?
Olace pairs your phone and other devices to the AI running on your desktop, another machine, or a cloud VM. On the same network they connect directly over your LAN; away from it, they reach each other through a live, secure WebSocket (WSS) relay. Every pairing is end-to-end encrypted between your devices, and your conversations sync across them, so you start on one and continue on another.
Does running Olace expose my computer to the internet?
No. Your desktop never opens a port or accepts incoming connections. Olace makes a single outbound connection to reach you, the way a browser does. Nearby devices connect directly over your network, and away from home the traffic passes through a WSS relay that is encrypted end to end and unreadable to Olace. No port forwarding, no firewall changes.
Does it work with Ollama and LM Studio?
Yes, both. Olace serves the models you already run in Ollama or LM Studio and leaves their local servers exactly as they are. Nothing about your current setup changes; your models simply become reachable from your other devices. More AI runtimes will be supported over time.
Do I need to deal with a VPN or Tailscale?
No. Olace reaches your desktop over your local network or its own encrypted relay, with nothing to set up. It works at the app level and changes no routing, so if you already run a VPN or Tailscale, both work side by side without conflict.
Why not just set this up myself?
You can. A VPN or SSH tunnel gets you a connection, and if you enjoy that, go for it. The catch is that it is something you build, secure, and keep maintaining, and it is easy to get wrong: an Ollama left open to the internet is a mistake people make all the time. Even done right, all you have is a connection to the machine. Olace is the secure path by default, encrypted and authenticated, with no public URL and nothing to set up, plus what a tunnel cannot give you: native apps on every device, simple pairing, and conversations that stay in sync.
Can Olace read my conversations?
No. We cannot read your local or paired conversations. Conversation sync across your devices is encrypted and zero-knowledge: only your devices hold the key, so we could not read it even if we wanted to. When you use a cloud feature, like web search or an open-source cloud model, it runs on our servers just long enough to get the response, then it is gone, never logged or stored. Those cloud models run on zero-data-retention providers that never keep or train on what you send.
What is Direct Mode?
Direct Mode is the strictest privacy setting, a hard guarantee that a conversation, in readable form, stays entirely on your own devices. It runs only on the AI models you have locally or on a paired device. Web access and external tools, like web search, research, and cloud models, are off in this mode. Conversations still sync to your other devices for continuity, always encrypted. Olace is private by default; Direct Mode is the switch that cuts off everything external.
What devices can I use it on?
Android, iPhone, Windows, macOS, Linux, and the web, all native apps. Pair any of them to the desktop, server, or cloud VM that runs your model.