Why Alto
The backplane under the tools you already use.
Work used to be something you sat in front of. Now it runs for hours without you, on machines you aren't at, started
by agents as often as by you. Something underneath has to keep it, where people and agents can both reach it.
That's a backplane: the board every card plugs into, and talks across. Keep your terminal, your agent apps and
your workflow. They plug into Alto.
Every session outlives the window, the app, updates and crashes.
Every session is machine:id, from any machine you have.
Every session says what it's doing, and when it needs you.
Any terminal, the app, your phone or an agent can plug in.
The terminal you use every day
iTerm2, Ghostty, Terminal, Windows Terminal, the VS Code terminal
What it's good at
It's yours: your font, your keys, your habits, years of them.
What Alto adds
Sessions that keep running when the window closes, the app updates or something crashes; every machine you have, one address away; and a mark on each session saying whether it's working, needs you or is done.
Together
alto new --attach starts a session right there, Ctrl-\ detaches, and alto attach studio:3 brings back any session from any machine. Alto takes no other key.
Logging in to other machines
ssh, remote shells, tunnels
What it's good at
A shell on any machine you can reach, with tools everyone knows.
What Alto adds
Sessions on every machine as machine:id, reachable from any terminal, the app or your phone, with no keys to copy and no ports to open: Tailscale on each device, and every device proves who it is. A slow link gets the current screen at once, not the backlog.
Together
Keep logging in to administer machines. Run the long work in Alto sessions, so a dropped connection never ends it.
Apps that run agents on tasks
a worktree and a branch per task, diffs and pull requests beside the chat
What it's good at
Planning, worktrees, diff review and pull requests in one window.
What Alto adds
Agents that keep running whatever happens to the app, on whichever of your machines fits, with a notification on your phone when one needs you.
Together
Keep it for planning and review. Start the agents that should outlive it in Alto sessions, from its own terminal; alto new --repo gives one its own checkout and branch, and alto fetch brings the branch back.
Agent managers in the terminal
a session and a worktree per agent, in a terminal app
What it's good at
A quick, keyboard-driven way to start and switch between agents.
What Alto adds
The manager and its agents keep running after your terminal closes, and you can reach them from your other machines and your phone.
Together
Start the manager in an Alto session (alto new --attach, then run it). It carries on exactly as before.
Boards and trackers that hand work to agents
an agent per card or issue
What it's good at
Deciding what the agents work on, and in what order.
What Alto adds
Somewhere for that work to run: the machine that fits, picked for you, in a session you can open, watch and answer.
Together
An agent picking up a card starts with alto new --in 'os:linux has:docker' --repo . and Alto picks the machine.
Agents in the cloud
a vendor's servers, started from chat or an issue
What it's good at
Nothing to install, and agents that run while your laptop sleeps.
What Alto adds
The same, on your own servers, VMs and homelab, with any agent: alto install makes any of them one of your machines, and your code stays where you put it.
Together
Use both: keep cloud agents where they suit, and run the rest on machines you own.
What it costs
Reaching other computers needs Tailscale on each. On Windows, a session can't yet move to a new build in place, so an
update means a new tab for it (Run again resumes agents). And it's early.
Kept deliberately small
No repo, worktree or browser panels. No planning or scheduling logic of its own: agents do that. No syncing with issue
trackers. Any of these comes only if real use shows a need.