Developers

Your agent gets a real computer — and keeps working after you close the laptop.

A dedicated cloud machine with a persistent filesystem, a real shell, and an agent living on it. Pull a repo, run it, ship it — from the browser, the desktop app, or SSH.

Where the day goes

01

The agent stops when you do

Local agent tools live and die with your laptop. Close the lid and the long task dies with it. Nothing runs while you sleep, commute, or sit in a meeting.

02

Your laptop holds every credential

A local agent runs as you, with your keys and your shell history in reach. Isolation means a container you have to build and maintain yourself.

03

Sandboxes with no shell — or shells with no memory

Hosted agents give you a scratch container that resets between runs. You can’t install anything that lasts, and yesterday’s work is gone.

04

Every repo needs onboarding, again

Dependencies, env vars, services. You did it once; the new hire does it again; the agent needs it done a third time before it’s useful.

What a machine changes

01

A persistent shell and filesystem

Real processes, real installs, real state. What the agent set up last week is still there this week.

02

SSH and a terminal UI

Drop into your machine from any terminal. A full agent client renders markdown, diffs, tool calls, and permission prompts — no browser needed.

03

Dev servers with preview URLs

Ports the agent opens are detected and surfaced as live links you can open straight from the session.

04

An editor next to the agent

A full VS Code in the browser, opened on the same checkout the agent is working in — for the edits you’d rather make by hand.

05

Work that runs unattended

Cron schedules, multi-step workflows, and webhook triggers. Start it at 6pm, read the result at 9am.

06

Your models, your keys

Bring your own provider keys, pick the model and agent per session. Keys resolve on the machine and never reach the browser.

Before you sign up

How is this different from a local coding agent?
Those run on your machine — your uptime, your CPU, your credentials — and stop when you close the laptop. Hoshi’s machine is always on, so the agent keeps going and you can reach it from anywhere.
How is this different from a cloud dev environment?
A cloud dev environment is a computer with nobody in it — you still drive every keystroke. Hoshi is that computer with an agent living on it, plus the org layer that distributes your team’s setup.
Do I lose my terminal?
No. SSH in and you get a real terminal client against the same machine, with the same sessions you see in the browser.
Can I use my own models?
Yes. Connect your own provider keys and choose the model per session; your organization can also provision providers centrally.