Control Center

See the workspace, tell Automous what to do, and stay in control.

Checking AI…
Desktop offline
Operating mode Assist AI can analyze shared screens, collect workflow data, and optionally operate a paired Windows desktop while approval gates remain enabled.
Automous workspace

Cloud browser + source capture

Cloud browser offline
AI
AI Cloud BrowserOpenAI Hosted Browser
Step-by-step live view
☁
Start a cloud browser session Automous will open Chromium in an OpenAI-hosted sandbox. This view refreshes after every AI browser action so you can watch the work happen step by step. Uses your existing OpenAI API key. The key must allow Agents API + Responses API access.
Browser not started No active session
SRC
Local source screenProcede / Excede
Source screen capture
Procede / Excede source Select the RO window once. Quick Snap or Auto Snap will keep reading it while you click through the RO.
AI reading source screen
No capture yetAuto Snap off
Automous AI

Command chat

A

I’m ready. Select a saved workflow, choose your Procede screen once, then Quick Snap each section as you click through the RO.

V0.09 adds a real OpenAI Hosted Browser session inside the app. Automous can operate the top browser while the bottom shared RO screen feeds it verified source data.
0 screens attached Assist
Saved workflow

RO Data Collector

0 / 0
Select a saved workflow to tell Automous what to pull from each Quick Snap.
Waiting for a Quick Snap.
Source evidence

Saved source images

0 kept
Only screenshots actually used as provenance for captured data are kept. The rolling 3-snapshot memory stays temporary.
Live execution

Task plan

Idle
✓
No task running Your next instruction will appear here as a step-by-step execution plan.
!
Review required before final actions Submission, deletion, payment, sending, and other consequential actions should require confirmation.
Saved procedures

Workflows

Save what Automous should collect, where each value belongs, and when it must stop for review.

Workflow profile

DTNA Warranty Claim

Extraction + mapping

Fields to collect

RequiredWhat to pullDestination field
Changes are saved in this browser.
Session log

Activity history

V0.09 stores saved workflows, collected RO field values, chat history, persisted source evidence, and cloud-browser session activity in this browser.

AI configuration

Settings

AI backendOpenAI API through Cloudflare Pages Function
Checking…
ModelConfigured server-side with OPENAI_MODEL
gpt-5.6-luna
Approval gatesRequire review before consequential actions
Enabled
Quick Snap + Auto SnapReuses the selected shared screen, can auto-snap every 5 seconds, and updates the active saved workflow
Enabled
Cloud browserChromium running inside an OpenAI-hosted Agents sandbox
Checking…
Browser runtimeNo Cloudflare Browser Run account or token required
OpenAI hosted
OpenAI Agents permissionsYour existing OPENAI_API_KEY must allow api.agents.read, api.agents.write, and api.responses.write. An unrestricted project key normally already has these permissions.
Hosted browser modelOPENAI_BROWSER_MODEL or default
gpt-6-astra
Computer-use modelOPENAI_COMPUTER_MODEL or default for optional local Windows mode
gpt-5.6-sol
API verificationRuns a live OpenAI Responses API + computer-tool request
Optional Windows Agent

Local desktop mode

Optional fallback for controlling Windows-only applications. The primary V0.09 workflow uses the OpenAI-hosted browser and does not require this agent.

Download Windows Agent
Not connected
Start the Windows Agent, copy its pairing code, then connect here.