Safety Controls
Doughy's safety controls sit between your AI team and the outside world. There is one rule, it applies to every account, and it is not a setting you have to find:
Anything that leaves your workspace waits for you.
What waits
A text to a tenant. An email to a lead. A phone call. A public listing. Anything that spends your money, and anything that can't be taken back.
Your AI team prepares all of it — writes the reply, gets the quote, lines up the vendor — and then stops and puts it in your approvals queue. You read it, and you decide. Every time.
There is no autonomy level to pick and no threshold to tune. You can't accidentally configure your way into an unattended send, because there is no configuration to get wrong.
What runs on its own
Work that stays inside your workspace runs the moment it's ready: reading what came in, drafting replies, triaging an urgent repair, updating a record, laying out the comps you brought, keeping your morning brief current.
The line is about consequence, not confidence. A draft that's wrong costs you ten seconds to fix. The same words sent to a real tenant at midnight cost you rather more. So drafting is instant and sending is yours.
Everything is logged either way.
The kill switch
The kill switch is real, and it's one tap away: Settings > AI Controls > Pause All Agents. When activated:
It's an emergency brake, not a preference. Use it if something unexpected happens, and take your time reviewing before re-enabling agents.
Daily budgets
Separately from all of the above, you set a daily spend limit and a daily call limit. These are cost controls, not permissions: when a day's budget is used up, your AI team stops until tomorrow.
The waiting rule caps what your AI team can *do*. The budgets cap what it can *cost*.
Where to find it
Settings > AI Controls holds the kill switch and your daily budgets. Anything that needs your sign-off lands in your approvals queue, one tap from any approval notification.
A dedicated console — one screen combining the live activity feed and cost views — is coming soon. Until it arrives, the Audit Log in the web app keeps the complete history of agent actions.
You can also ask your AI team "What's waiting on me?" by voice, or have your daily brief — delivered to the channel you choose, at the hour you choose — include what's pending.