Early accessShowcase →Demos →
People and AI agents.One operating platform.Any agent, on your machine.
At work in your terminal, browser and desktop apps. You choose what each one may touch, and answer it from your desk, browser or phone.
Your subscriptions. Another cloud bill.
Only your access and launch news. To leave, write to contact@cognix.ws.
- ComputerActs on your desktop the way you would: any window, the clipboard, any program. Only for agents you allow.Read the docs →
- BrowserDrives a browser tab beside the agent. You see the same page and can take over at any time.Read the docs →
- InboxAgents file notes and questions in one inbox. A question waits there for your answer.Read the docs →
- ChecklistThe agent's plan for the task, step by step, live on its card.Read the docs →
- MessagesAny agent can message any other, in any workspace, and wait for the reply.Read the docs →
- AgentsAn agent can start helpers with a task and get their answers back.Read the docs →
- ArtifactsAgents publish pages and documents you see beside them, with every version kept.Read the docs →
- ExtensionsBuild your own modules: new tools, panels and integrations for your workspace.Read the docs →
- LifecycleInstall, update and remove each CLI from the app (or open its vendor's uninstall guide), and run its checks when you ask.Read the docs →
- LaunchChoose how each CLI starts: executable, arguments, environment, account, saved templates. See what COGNIX[WS] adds before it runs.Read the docs →
- Sign-inSign in to every CLI from the app and keep several accounts. Keys you save here are encrypted on your computer.Read the docs →
- PermissionsAn agent gets only the COGNIX[WS] tools you tick when it starts, Computer off until you choose it. Helpers never get a tool their creator lacks.Read the docs →
- ResilienceAgents keep working when you close the window or COGNIX[WS] restarts. After a reboot, resume each one with a click; leftovers are cleaned up.Read the docs →
- SessionsPick up the exact conversation after a stop, a restart or a crash. Fork it into a second agent, or continue it in another CLI.Read the docs →
- ObserveSee which agent is working, needs you or is ready, and what it costs, split between API and plans.Read the docs →
- ContextSee which instructions each CLI reads, and edit its skills, memory and settings in the CLI's own files.Read the docs →
- ConnectorsManage each CLI's MCP servers and plugins from the app, with live status where the CLI reports it.Read the docs →
- PhonePair your phone with a QR code and keep working from it: answer agents, open a terminal, edit files.Read the docs →
- AutomationsRun an agent, a script or an Assistant on a schedule, from a webhook or with one click. Agent runs and every failure land in your Inbox.Read the docs →
- RemindersA time you must not miss: on a pin, on its own, or proposed by an agent. The card stays on screen until you act, wherever COGNIX[WS] is open.Read the docs →
- AnalyticsSpend, activity and plan limits across your CLIs: usage from each CLI's own files, limits from the provider you signed in to.Read the docs →
- Local modelsRun open models with llama.cpp: download and load them from the app, and use them from Pi, Omp and your Assistant.Read the docs →
- SnippetsSave prompts and commands once, with blanks to fill, and send them to an agent, a terminal or your Assistant.Read the docs →
- WebhooksSend COGNIX[WS]'s events to your own services, signed, in order and retried.Read the docs →
- DevicesSee the browsers, phones and tokens paired with your COGNIX[WS], and forget any of them.Read the docs →
- BackupSnapshots of COGNIX[WS]'s data to a folder you choose, on the schedule you turn on, with encrypted copies elsewhere and restore from the app.Read the docs →