Answer Radar / Brief and schedules

A weekly brief that says when nothing changed, and the schedule that feeds it

Set a project to re-measure weekly or monthly. Then get one email per window comparing it with the window before, per engine, with the queue of things to act on. When the week was quiet, the brief is short and says why, instead of manufacturing a highlight.

Start tracking Turning on briefs

Updated 2026-09-04

What the brief and the schedule are

Two separate switches. One measures, one reports.

A schedule runs a project's audit on a cadence without you pressing the button, spending credits like a manual run and obeying the same caps. The brief is the recurring decision for your account: what moved across every active project since the last window, which queued actions are outstanding and which are new, and what was completed or dismissed. It is off until you turn it on, weekly by default, monthly for quieter accounts, and it returns nothing_meaningful_changed with the reason when all three are quiet.

Our own brief, pulled on 4 September 2026

Read-only call on our account. Two projects; the linkeddit.com project had one run in the window and none before it, which is what the movement block reports.

answer_digest_get({ days: 7 })

answer_schedule_set({
  project_id: "<your project id>",
  cadence: "weekly",
  timezone: "America/New_York",
  enabled: true,
  idempotency_key: "schedule-2026-09-04"
})

The digest call is real and its response is below. The schedule call is shown for its shape and was not executed, because it writes a schedule on our production project.

answer_digest_get, days 7, abridged

version: digest-v1
nothing_meaningful_changed: false. Nine cards were raised inside the window.
new_action_count: 9 new, 9 open, 0 completed, 0 dismissed
period: 29 August to 5 September 2026 against 22 to 29 August. The two windows never share an instant.
movement: not_comparable. Five engine segments (Claude, Gemini, ChatGPT, two Perplexity instruments), every one absent_in_baseline: comparable 0, not_comparable 5. Threshold basis max(0.10, 1/samples) per segment. confidence null.
queued, top 5 of 9: Not represented for: competitor monitoring software for tracking pricing changes and product launches (rank 1); Crayon alternatives (2); competitive intelligence tools for B2B SaaS (3); competitor analysis tool for product marketers (4); answer engine optimization tools (5).
second project: No brand name, no segments, no cards: nothing_meaningful_changed true.

From the response's own notes: a calendar window targets nothing, so control prompts are unavailable and ambient_change_not_isolated is unreachable here. The bound was not applied, not passed. No sentence explaining what any of this means is written; the facts are yours to narrate.

The three parts of a brief

Per the answer_digest_get contract, which the email is rendered from.

Movement

What changed across every active project in the last window versus the equal window before it, per engine segment, with each number's disclosure. A window that cannot be compared is not_comparable, never a fall to zero.

Actions

Which queued cards are outstanding, which are new since the last brief, and which were completed or dismissed. An old open card is listed but does not make the week eventful; new_action_count is what changed.

The quiet verdict

nothing_meaningful_changed is true only when all three inputs are quiet: movement held still or could not be compared, no card was raised, nothing was closed. Then the brief is short and says why.

How schedules behave

From the answer_schedule_set contract and the schedules help article. The one question schedules always raise is the last row.

Rule	What happens
Cadences	Weekly or monthly, in the IANA timezone you name. Daily exists but is restricted to internal accounts; the response lists what your account may pick.
Enabling	Starts no run now. The first fires one cadence from now, at the next_run_at in the response, in your local hour, preserved across daylight-saving changes.
Pausing	enabled false on its own. The cadence and timezone are kept, so resuming is one call with enabled true and does not fire the instant it comes back.
Cost and caps	A scheduled run reserves budget first and is refused by the same per-run and daily caps as a manual one. Missed slots are skipped, never replayed as a backlog.
Plan changes	A subscription that stops including Answer Radar pauses every schedule on the account. Winning it back does not auto-resume them; you resume the ones you want.
Why did mine not run	answer_activity_list records a skipped slot as a schedule.dispatch action with outcome denied and, in evidence mode, the reason and the slot it was for.

What this is not

Things the alert-style pages in this category imply, and where we stand on each.

Not real-time alerting. Engines are measured on a cadence you set, because every sample costs a credit and a continuously polled engine is somebody's bill.
Not a written summary. The email is fixed sentences chosen by fixed rules, with no adjective that varies with the data. Narrative is your assistant's job, reading the same payload.
Not on by default. Your toggle, then an operator switch, then an unsubscribe link in every mail.
Not a backlog. A missed schedule slot is skipped, not replayed, and a crash between claiming a brief slot and sending loses one brief rather than risking a duplicate.
Not a confidence score. None is computed, so none is printed.

Frequently asked questions

Does Answer Radar send real-time alerts when an AI mentions a competitor?+

No. Answer engines are measured by running your prompt set against them, which costs credits per sample, so the product measures on a cadence you set, weekly or monthly, and briefs you on the comparison between windows. A tool promising real-time alerts on AI answers is either polling the engines continuously at someone's expense or describing something else. We would rather tell you the cadence than sell you the word alert.

What is in the weekly brief?+

Three things: movement across every active project in the last window versus the equal window before it, per engine segment; which queued actions are outstanding, new, completed or dismissed; and a verdict of nothing_meaningful_changed with the reason when all three are quiet. Every number carries its disclosure. There is no confidence value and no written summary sentence.

Why is the brief off by default?+

Because nobody should be mailed a report they did not ask for. Only your own toggle enables it, sending is also gated by an operator switch that is off unless turned on, and every email carries an unsubscribe link. There is deliberately no path that enables briefs for everyone.

Why does your own brief say not_comparable?+

Because our project had one audit run in the last week and none in the week before, so every segment reports absent_in_baseline. That is the honest answer for a project measured once: nothing was measured against anything. It is a different answer from no_significant_change, which means two windows were compared and nothing moved.

Does turning on the brief start audits?+

No. The brief reads what audits, scheduled or manual, have already measured. Schedules are set separately per project with answer_schedule_set, and enabling a schedule does not start a run now either; the first fires one cadence later.

Who writes the summary?+

You do, or your assistant does. The email is template-thin by design: every sentence is a fixed string chosen by a fixed rule, and no adjective varies with the data. Through the connector, answer_digest_get returns the same payload with your own days and cadence, and an assistant reading it can write a better narrative than any template we could ship.

Measure on a cadence, get briefed on the difference

Schedules and the brief are part of Answer Radar, included with the Compete plan, in the browser and through the MCP connector.

See pricing How Answer Radar works