Roadmap
Shipping in the open.
Everything below is the plan we work from. No dates — an item moves left when it is real on real machines, not when a deck says so.
In beta now
- Session archive across 45 formatsClaude Code, Codex, Gemini, Grok, Kimi, Qwen and more, in one local library.
- Usage meters and Needs You flagsOne meter across nine quota systems, and the moment an agent is stuck.
- Local voice on your own machineDictation and speech that never leave the machine.
- Signed installers with auto-updateOffline install, signed binaries, updates that land on their own.
- Automater Desktop betaThe full workspace, with Lite as its tray.
Next
- Live Visualizer for every repoWatch agents work on a live map of the project.
- Resource monitor hog scoresLive CPU, memory and I/O per running AI app.
- Pop-out mini-apps on every platformAny widget becomes its own always-on window.
Later
- Bots with their own computersLocal, self-hosted or cloud Linux desktops an agent can drive.
- Fleet factory runs across machinesOne job fanned out over every box you own.
- Shared agent computers for a teamA pool of desktops the whole team can hand work to.
- Agent gatewayAny LLM can read job status and steer work that is already running.
Beta means it runs on real machines now and can still change. Tell us what is missing.
Coming to Automater ProPro
Take back your life.Tell Auto. It routes.
One chat box in the tray. Auto sends your request to the first engine in your priority order that is up and under quota, so you stop picking models and hopping apps.
- Your order, not a guess.
- Claude, Codex, Grok, Kimi, Qwen, Antigravity, Z.ai and Muse, in the order you set.
- Quota-aware.
- An engine at 90 percent of its quota is benched until it recovers.
- Pin when you mean it.
- Name an engine in the message, or pin one in the card, and Auto steps aside.