Autopilot
Autopilot prepares your outreach overnight so you wake up to a stack of ready-to-review drafts. It never sends anything on its own — you always review and send.
How it works
Each night, autopilot tops up your draft queue to the number you choose ("Keep this many drafts ready"). As you send or clear drafts during the day, the next night's run refills the queue back to your target.
It decides who gets the next message in priority order:
-
Live conversations — contacts you've already reached who are due a follow-up, starting with whoever has waited longest since your last message. That way nobody you've started talking to is left hanging while newer conversations get the attention. Each follow-up sticks with the same problem the conversation opened on and brings one new thing to it each time — a fresh angle, a proof point, a pointed question — so nobody gets the same message twice. If a channel is exhausted (for example a LinkedIn request that hasn't been accepted yet), autopilot uses the next available channel, such as email — the same order you set in Channels.
If someone replies while a follow-up is waiting for you to review — even while your AI Agent is still writing it — that draft is dropped, so nothing outdated goes out to someone who has already spoken up. Nothing new is drafted for them on its own: open their conversation and ask your AI Agent to write a reply to what they said.
-
Ready to contact — contacts staged for outreach that don't have a draft yet.
-
Qualified — qualified contacts get moved into your outreach flow so their first message is drafted.
-
People not qualified yet — these are qualified first, and if they qualify their message is drafted on a later run. Autopilot starts with your Possible Fit contacts, the ones your AI Agent's screening found promising, most promising first. It only turns to New contacts when no Possible Fit is waiting, again most promising first, and it leaves out anyone screening already ruled out. That way your credits go to the people most likely to be a fit.
Autopilot follows your team's Automatic AI work settings: whether newly qualified people are researched, and whether older research is refreshed before a message is written.
Autopilot stops once your queue is full, so it never floods your review list. It only qualifies new contacts when it ran out of people to write to, and only as many as it needs to fill the gap. That keeps your credits for the people you can actually message. Each night it also stays inside your nightly credit budget, and drafts for people already in your pipeline are free; see How much autopilot spends each night.
Turning it on
Autopilot is on by default. Manage it under Account → Outreach Settings, per team:
- Autopilot toggle — turn the nightly fill on or off.
- Keep this many drafts ready — how many drafts you want waiting each morning, from 1 to 200. Pick the number you will actually review; autopilot stops writing once that many are waiting. If the number looks lower than you remember, raise it here. Next to it you see what a night at that number costs in credits, and tonight's budget.
It uses the same channel preferences and cadence you configure for each channel, so it respects your per-contact limits and spacing.
Want the next batch now?
Autopilot tops your queue up and then waits for you, so it won't add more until tomorrow morning. If you've cleared the queue and want to keep going, Run autopilot now appears on your Tasks page when the queue is empty. It runs the whole cycle immediately, and drafts appear as they're written.
There's a short cooldown afterwards, so pressing it twice won't start two runs.
You can also just ask: open your AI Agent anywhere in Doers and say "find me some more people to reach out to". It starts the same run, for your queue only. There's a short cooldown either way, so asking twice won't start two runs.
What autopilot never does
- It never sends a message — every message is a draft awaiting your review.
- It never exceeds your "drafts ready" target.
- It never messages a contact who already has a draft waiting.
- It never drafts for someone you've skipped, until the date you picked has passed.
- It never leaves a stale follow-up in your queue — a reply from the contact drops the waiting draft.