The problem this solves
AI answers move without announcement. A model update ships, a competitor publishes something citable, and a question you comfortably owned starts recommending someone else - and nothing you monitor today will tell you. Teams typically discover a displacement weeks later, from a prospect who mentions what ChatGPT told them, by which point the new answer has hardened and the sales team has already felt it in calls without knowing why.
Weekly dashboards help, but a scoreboard read on Friday still buries the event: what you need is the delta itself, the day it happens, with the evidence attached. Displacement is rare enough that no person will keep checking for it week after week, and consequential enough that it should interrupt a Slack channel the moment it occurs. That mismatch - too rare to staff, too costly to miss - is the exact shape of work a standing mission exists for, and the diff-against-archive mechanic is what turns repeated sampling into genuine detection instead of another dashboard.
How the mission runs
- Set the tripwires. The standing mission locks your tracked questions and the brands worth alerting on - you and the rivals you care about. Every run compares against the stored prior state, which is what turns sampling into detection.
- Sweep on cadence. Each week the mission re-samples every tracked question across the engines and diffs the answers against the archive: new brands mentioned, citations gained, citations lost, recommendation order changed.
- Alert only on real movement. Wobble is filtered - engines rephrase constantly - and only durable changes fire: a competitor newly cited, your page dropped, a lead recommendation flipped. Silence means nothing moved, and silence is the usual, correct output.
- Ping with the evidence and a move. The Slack alert carries the question, the before-and-after answer excerpts, what specifically changed, and one recommended response - refresh the displaced page, publish against the gap, or watch one more cycle before acting.
The prompt
This is the exact objective the agent receives. Swap the obvious placeholders for your own domain, segment or channel and run it as-is from the console, Slack, or the API.
What comes back
A standing mission that watches your citation ground and speaks only when it shifts: Slack alerts carrying the captured before-and-after answers, the specific change, and one recommended response, plus a growing archive that makes every displacement traceable to the week it happened. The alert usually arrives while the displacing content is days old - which is when a refreshed page or a fast rebuttal post still has a real chance of taking the answer back before it settles.
Make it yours
- Add an escalation rule: a displacement on one of your five money questions goes to the leadership channel, the long tail stays with the working team.
- Track a launch: add the questions a new product must win and watch citations arrive as your pages get picked up.
- Run it daily during a rival launch window, then drop back to weekly when the field settles.
Frequently asked questions
How does it avoid crying wolf?
Answers naturally vary between samples, so the mission distinguishes phrasing wobble from durable change and holds single-run movements until they persist. The alert threshold is part of the mission text, so tightening or loosening it is a one-line edit.
What does a quiet week look like?
No message. The run still happens, the archive still grows, and the Schedules tab shows the completed run - but the channel only hears about movement. A weekly all-clear line is available as a variation if silence makes the team nervous.
Does this need a particular plan?
Standing missions are a Growth and Scale capability, and this playbook is a schedule by nature. On Starter, run it manually as a one-off sweep whenever you want a displacement check; the diff against the previous run still works.