Autonomy controls

// the station keeps working while you're away — on your leash

The autonomy model

Autonomy in bdibranding is a dial, not a switch: you set initiative (how proactive agents may be) and reach (what they may touch) explicitly. Nothing self-escalates. The honest pitch: bdibranding is not an "autopilot for your life" — you direct; it executes and proposes.

Night Shift

When you step away, Night Shift takes over: agents pick up standing goals, continue queued work, and explore ideas — inside the leash you configured. Every Night Shift run is jailed: real deliverables are produced in a sandboxed workshop area, and touching anything outside it requires a grant you gave beforehand. A paired messaging account is different: its direct messages run shell commands, your connected services and desktop control without asking, even where the station window would ask first, unless you send /approvals on in that chat or turn it off in SETTINGS › PERMISSIONS.

Stopping work

Use Stop in COMMS (or /stop) to interrupt the current run. To stop everything at once — every agent, messaging reply, routine, loop, the night shift, delegated workers and background terminals — type /halt in COMMS. It asks first: type /halt again, or /halt confirm, within 30 seconds, or /halt cancel to back out. In the desktop app the tray item Pause Automation (E-STOP) does the same, even with the window closed. Routines, loops, the night shift and background learning then stay paused until you press RESUME AUTOMATION in the station window. You can also pause routines or loops in their own panels, and set Night Shift initiative to WAIT to prevent new away-work. Recovery preserves your existing permissions and job settings.

A paired owner can also stop all station work from the phone: send /halt in the direct message with the bot, then /halt confirm within 60 seconds. RESUME AUTOMATION is only in the station window.

Honest reporting

When you come back you get a digest of what actually happened: which runs executed, what they produced, what they cost, and what was declined or blocked. Deliverables open with one click. Work that landed while you were watching is shown as it lands — bdibranding distinguishes "shown to you" from "done while nobody watched" and never blurs the two.

Schedules

Cron-style routines run real agent sessions on a timer — a morning brief, a nightly digest, a weekly cleanup. Every scheduled run surfaces as a readable session afterwards: same transcripts, same ledger, same honesty as attended work. Scheduled runs can notify you through your connected channels when they finish.

Quests

The quest system turns standing goals into visible, refreshing work the station proposes and tracks. Quests are generated from your actual dossier and interests — and like everything else, a quest card only ever reflects real backing state.