You decide how muchthey do alone.
One dial for every agent and every kind of action. Turn it up as your trust grows, and turn it down in a second.
From off to full autopilot.
Choose a level for each agent and each kind of action.
Off
The agent does not do this action. Nothing is drafted, sent or changed until you switch it on.
Propose
Drafts go to your queue and you approve each one. Nothing leaves without your yes.
Auto within rules
The agent acts alone when every rule you set passes. Anything outside the rules goes to your queue instead.
Full autopilot
The agent chooses targets, timing and follow-ups itself, inside your caps and the safety rules. You get a digest, and you can undo or stop at any moment.
- Opt-out and address on every email.
- Never-email list that no agent can override.
- No duplicate sends to the same person.
- Spend caps and automatic pauses on bounces or complaints.
- Kill switches, an undo window and a decision journal.
One agent, different levels.
An example setup. Relay can sort replies on its own while every answer it writes still waits for you.
| Agent | Action | Level |
|---|---|---|
| Scout | Find and score prospects | Full autopilot |
| Scout | Send the first email | Propose |
| Relay | Sort incoming replies | Full autopilot |
| Relay | Send answers to interested prospects | Propose |
| Relay | Reply to "not now" messages | Auto within rules |
| Quill | Publish an article | Propose |
| Lens | Change a live page | Off |
| Echo | Schedule approved posts | Auto within rules |
Example only. Your settings are your own, and you can change any of them at any time.
Limits, pauses and a way back.
Spend caps
Set a daily cap per agent and for the workspace. Every AI call is metered, and work pauses when a cap is reached.
Automatic pauses
Sending stops by itself when the bounce rate passes your limit or a complaint comes in, and Chief tells you why.
Undo window
Actions taken without you can be undone for a while afterwards.
Kill switches
Stop one agent, one project or the whole crew in a second.
Decision journal
Every autonomous decision is recorded with what was done and why, so you can always check.
Digest and reports
A short daily digest from Chief and a Monday report from Pulse keep you informed without opening the app.
How trust grows.
There is no leap of faith. Hand over one kind of action at a time.
Start on Propose
Every draft, reply and proposal waits for your yes. You see exactly how each agent behaves.
Read the journal
After a few days you can see which kinds of action the crew gets right every time.
Raise one level for one action
Move that one kind of action to Auto within rules. Keep everything else on approval until it earns the same trust.
About control.
Can I set different levels for different agents?
Yes. The level is set per agent and per kind of action, so Relay can sort replies on its own while every answer it writes still waits for you.
What does Auto within rules mean?
The agent acts alone when every rule you set passes. Anything outside the rules goes to your queue instead of being sent.
Is there anything I cannot switch off?
Yes. The safety rules run at every level: an opt-out and address on every email, a never-email list no agent can override, no duplicate sends, spend caps and automatic pauses.
What if an agent does something I did not expect?
Stop it with its kill switch, undo the action if it is inside the undo window, and read the decision journal to see why it acted.
Stay in charge. Get the work done.
Give BrainPilot your website and set the pace.