Campaigns & autopilot
A campaign is a recurring content program: a goal, a narrative arc, and a cadence that keeps generating posts until you stop it. It is the difference between ReachPill and a one-asset-at-a-time tool.
Creating a campaign
- 01
Describe the goal in plain words
“Launch week for our new analytics feature, aimed at heads of growth” is a complete input. The AI builder does the structuring.
- 02
Review the plan
The builder proposes a narrative arc (what each post contributes), a cadence (daily, weekdays, 3× a week…), the target platforms, and a credit cost estimate.
- 03
Commit
Nothing is generated and nothing is billed until you accept the plan. After that the queue fills itself on schedule.
Review mode vs autopilot
Approval is a per-campaign choice:
- •Review required — every occurrence waits in the queue (and in Slack, if connected) until a human approves it.
- •Autopilot — occurrences publish on schedule without manual review. You can still watch everything in the queue and calendar.
The queue
Each occurrence carries its per-platform variants — square for X, 16:9 for LinkedIn, 4:5 for Instagram — with captions re-toned per channel. Actions:
- •
Approve— ships on its scheduled slot. - •
Reject & regenerate— produces a fresh version grounded in the same brief. - •
Skip— drops the occurrence without breaking the arc. - •
Reschedule— moves it to a better slot.
Topics let you steer what upcoming occurrences are about without editing individual posts, and the Calendar view shows the whole month read-only.
On-brand verification
Every finished post is checked against your Brand DNA before it reaches you — arithmetic, not vibes: captions are matched against your banned terms, and the image's dominant colors are measured against your declared palette. Problems show as chips on the card ("banned term: cutting-edge", "palette drift"); clean posts show nothing. A banned term is a hard fail: autopilot will not publish it — the post is routed to human review instead.
It learns from your reviews — with your approval
Recurring review feedback ("too salesy", "we don't say leverage") is mined into suggested DNA updates— terminology rules, do's & don'ts, voice examples, facts — that appear in an inbox on the DNA page. Accept applies the change (versioned, so History can undo it); Dismiss closes it. Nothing ever changes itself.