Install on your lab’s Macs.
Download the signed, Apple-notarized installer. Install it, then open Edvi Worker from Applications.
Apple Silicon · macOS 26+PRIVATE COMPUTE FOR RESEARCH LABS
A simple way to build a private network of Macs for your research lab. Connect the hardware you already own, grow your AI compute capacity, and put more experiments to work in parallel.
Apple Silicon · macOS 26 or later · Early access
Windows & Linux — coming soon.
FROM LAB MACS TO SHARED CAPACITY
No Python installation. No Terminal setup.
The installer brings the runtime with it.
Download the signed, Apple-notarized installer. Install it, then open Edvi Worker from Applications.
Apple Silicon · macOS 26+Create an account, confirm your email, and sign in. Edvi downloads a small AI model and connects your Mac.
Use the same account on each MacOpen your dashboard, send a prompt, and read the result. Each job runs on one available Mac in your account.
Add another Mac whenever you’re readyA FOUNDATION FOR DEMANDING AI RESEARCH
Edvi’s direction is simple: make a network of Macs an accessible platform for compute-intensive AI workloads, research experiments, and agents.
Use multiple Macs to process independent inference jobs in parallel. Today’s early release starts with prompt-based jobs.
Add another Mac to increase the number of jobs your network can handle. Each job uses one Mac; memory is not pooled across machines.
Agent execution and broader research workflows are the direction we’re building toward. They are not included in the current early release.
BUILT AROUND YOUR LAB’S HARDWARE
Build a private compute pool from your lab’s Macs. Add machines to run more independent AI jobs at once, without setting up each experiment on a separate computer.
See connected workers, submit prompts, and check job results from your browser.
Edvi starts when you log in to your Mac. Pause or resume your worker in setup, or disconnect it from your dashboard.
The coordinator sends a job to a connected Mac, where the model generates its answer. Prompts and results pass through the hosted service.
GOOD TO KNOW
This installer supports Apple Silicon Macs (M-series chips) running macOS 26 or later. Keep enough free storage for the app and model download, and memory available for inference.
Your worker pool consists of the Macs connected to your account, rather than a public pool of other people’s computers. Models run on those Macs. Coordination is hosted: prompts and results pass through the Edvi service, so this release is not a fully on-premises or air-gapped deployment.
The current release supports prompt-based inference, with each job running on one Mac. Adding Macs increases parallel job capacity; it does not combine their memory or split a model across machines. Agent execution and broader compute-intensive AI workflows are planned directions.
Windows and Linux support is coming soon. The current installer is for Apple Silicon Macs.
No. One Mac is enough to run your first prompt. Sign in with the same account on another Mac to add it to your dashboard. A single prompt runs on one Mac; its model is not split across machines.
Create an account, verify the email link, then sign in to connect your Mac. The model downloads automatically. Keep your Mac awake and connected to the internet until setup says Ready.
Keep your Mac awake and online while it processes jobs. Login startup runs after you sign into macOS. You can pause the worker whenever you need its resources.
Yes. Edvi is in early access. The installer is signed and notarized, but automatic app updates and password recovery are not available yet. Keep your account password safe and install new releases manually.
You can create one in Mac setup or on the Edvi dashboard. Confirm the link in your email before signing in. Existing users can simply sign in.
START BUILDING YOUR LAB’S NETWORK
Install Edvi on your Macs. Connect your private worker pool. Run your first AI job.