Autopilot and workflows

Everything that acts on your brand's behalf lives under one menu, Actions › Automation. Four parts are explained here: the Autopilot that carries out approved work under a risk policy you set, Workflows you assemble from steps on a canvas, the Agent that checks your rules every night and proposes work, and Campaigns that tie work to one measurable goal. The other two tabs, Tasks and Digital PR, are covered in Sources, outreach and PR.

#Where to find it

The autopilot policy is set in the user menu (your name at the bottom of the sidebar) under Settings, on the Autopilot tab.

#Autopilot

#Risk tiers

Every piece of work the autopilot can carry out belongs to one risk tier:

• Automatic — done at the next pass.
• After 24 hours (you can veto) — you are told, and have 24 hours to cancel before it runs.
• Wait for my approval — waits in the queue until a person approves it.
• Off — never done.

Someone else's platform is never automated, whatever you choose: the system prepares the text, you send it.

#Daily limits

Under Daily limits you cap how many Emails, Posts (Google Business Profile updates) and Publishes may go out per day. The defaults are 20 emails, 5 posts and 10 publishes.

#Setting the policy

1. Open Settings › Autopilot.
2. Choose a mode for each tier and adjust the daily limits.
3. Press Save.

#The queue

The Autopilot tab lists what the system will do on your behalf, with each item's Risk tier, Status and when it runs (Runs:). Statuses are Awaiting approval, Scheduled, Running, Done, Failed and Cancelled.

• Approve a waiting item to schedule it.
• Cancel to drop it — this is also how you veto an item in its 24-hour window.
• Retry a failed item.

Approved content, ready-to-send outreach messages and profile updates all land here.

#Workflows

A workflow is a chain of steps — measure → check → act → notify — that runs by hand, on a schedule, or when something happens.

#Creating one

1. On the Workflows tab, under New workflow, enter a Name and pick a Template, or Empty canvas.
2. Press Create. The full-screen builder opens.
3. Add steps from the palette on the left: click one to add it after the selected step, or drag it onto the canvas. The Power Steps tab inserts a whole ready-made chain.
4. Connect steps. A connection means "runs after"; a connection that would create a loop is refused.
5. Select a step to edit its Parameters in the Details panel, which also shows what the step Requires and Produces.
6. Click the trigger node to choose when the workflow runs, and switch Enabled on. When it is off, the workflow runs only by hand.
7. Press Save (⌘S). Save validates the graph; steps with errors are marked and nothing is saved until they are fixed.
8. Press Run to try it now. The log appears in the Runs panel.

#Triggers

An event can start a given workflow at most once per day. If a scheduled run is missed (for example during maintenance), it is caught up once within 24 hours; older slots are skipped rather than run in a burst.

#Steps

A check step stops the run cleanly when there is nothing to act on. Steps that use an outside tool (Search Console, Semrush, Ahrefs, Moz, a CMS, Notion, Airtable, Asana, ClickUp, Google Docs, Slack) need that tool connected under Integrations. Publish to CMS has an As draft switch: on, the post stays a draft in your CMS.

#Templates

Templates include Gap → brief, Stale page → refresh, Citation loss → refresh + Slack, Weak format → audit + actions, Coverage gap → prompt ideas, Visibility watch (daily), Rival watch (daily), Source opportunity → notify, GSC drop → refresh + Notion, High-volume gap → brief → CMS draft and Authoritative source → outreach + Asana. Start from one and change the steps.

#Checking and debugging

• The header shows No issues or N issues. The builder warns when there is no check step (the workflow would act on empty findings) or no notify step (it would end silently). QA workflow in the copilot runs the same check.
• The Runs panel shows each run's status — queued, running, done, stopped, failed or skipped — and what every step found. Resume from failure restarts a failed run from the step that failed.
• Copilot explains the workflow, says which measurement triggers what, and suggests up to three changes from your numbers. It never edits the canvas itself.

#Agent

The Agent tab holds playbooks: rules the agent checks every night at 06:30 UTC, after the day's jobs finish. When a rule's condition is met, it prepares a proposal and waits for you — nothing is published without approval.

#Creating a playbook

1. Under New playbook, enter a Name.
2. Type thresholds in your own numbers, or leave them at the defaults. The thresholds are Visibility drop threshold, Citation loss threshold, Stale page threshold, Coverage gap threshold, Weak format threshold, Competitor surge threshold, New gap threshold, Refresh confidence and Source impact threshold.
3. Tick the Permitted kinds of work: Draft a brief, Draft content, Open a source task or Suggest only. Briefs and content are generated after approval (paid); a source task opens directly; a suggestion generates nothing.
4. Press Create playbook. Use Disable / Enable to switch it off and on.

#Approving proposals

Proposals appear under Waiting for approval with the reason in plain numbers. Approve queues the work; Reject asks why. The Decision log keeps the history, and Acceptance rate shows how often you agree. The agent learns from your decisions: after repeated rejections it raises the threshold that produced them, and acceptances lower it slightly. The numbers stay visible, so you can set them back by hand.

The agent stays quiet on rates built from fewer than 4 measurements. Approving starts the work; publishing still needs you to approve the content separately.

#Campaigns

A campaign is one goal, a fixed set of prompts and the work accepted toward it, with before → now → target on one line.

1. Press New campaign and enter a Campaign name.
2. Choose a Metric: Visibility (share of answers naming the brand), Citations (answers citing your own page) or Share of voice. Set a Target — a percentage, or a number of answers for citations — and optionally a Deadline.
3. Set the scope with Scope: topic and/or Scope: intent; leave both empty for all active prompts.
4. Choose the Kinds of work accepted (content gap, comparison page, answer format, refresh, source/outreach, directory/listing, community, Wikipedia).
5. Press Start campaign. The baseline is taken now, and the current version of your brand vault is pinned to the campaign.
6. Under Opportunities, select open findings that fit and press Accept selected. Accepted items show as planned on the action plan.

The scope stays open — new matching prompts join — until the first accepted item goes live or you press Lock scope; from then on the prompt list no longer changes. Finish campaign closes it.

The list shows Running campaigns, Target reached, At risk (past due, or will not make it at this pace) and Work completed. With a deadline, the pace column projects where the campaign will be by then at the pace so far; without one, it does not guess. A campaign's detail page shows Baseline, Now (last 30 days), Target, a daily trend of the scope and each prompt's mention rate. A figure built on too few answers is marked sample too small.

#Limits and plans

See Plans and limits.

#Common questions

Nothing ever reaches "Done". Check the policy in Settings › Autopilot: by default every tier waits for approval, so items sit in Awaiting approval until you press Approve.

Why is my email still waiting even though I set "Automatic"? Your plan does not include outward autopilot, so own-account work falls back to approval.

My workflow stopped after the first step. A check step found nothing to act on — that is a clean stop, not a failure. Open the run in Runs to see what each step found.

#Related

• Action plan
• Content
• Sources, outreach and PR
• Alerts and inbox
• Integrations