Skip to content
RookAtlas
On every plan, Free included

Hear about real drops, not noise

A drop alert fires only when the paired test of this run against the last one, on the prompts both asked on the same engines, puts the fall beyond run-to-run noise. Everything else waits for a Monday summary, in Slack or your inbox.

  • All eight AI engines on every paid prompt run
  • A 95% interval on visibility and citation rates
  • Every run priced before it starts

Opening soon: sign-up opens with live measurement, after our provider pilot.

A drop alert in Slack
jbl-ai-visibility
RookAtlas Mon 06:42

JBL: visibility dropped

30% (20–43%) on the latest run, against 57% (44–69%) on the run before, same eight engines. On the prompts both runs asked, the paired test puts the fall beyond run-to-run noise; if the next run agrees, it's a change.

Open the run See the prompts

RookAtlas sends a brand's alerts to its own Slack channel and, for each person who turns them on, by email: a run finished or failed, a visibility drop the paired test supports, against the previous run on the same engines and the prompts both asked, and a Monday summary of the week with the weakest engine, new gaps and searches you don't track yet. Signed webhooks carry runs, actions and published articles to your own systems.

Measured on eight engines

  • ChatGPT
  • Gemini
  • Perplexity
  • AI Overviews
  • AI Mode
  • Claude
  • Grok
  • Muse
01

A drop means a drop

RookAtlas compares this run with the last one on the prompts both asked on the same engines, each counted once. Only when the 95% interval of that change lies below zero, with at least 10 such prompts, does an alert fire; a smaller or untested move waits for the summary.

  • Same engines compared on both runs
  • A paired 95% interval below zero
  • Untested moves called observed in the summary
The drop rule
Could be run-to-run noiseQuietBefore53%Now46%On the prompts both runs asked on the same engines, the change's 95% interval includes zero, so nothing is sent.
Drop alert sentAlertBefore57%Now30%On the prompts both runs asked on the same engines, the change's 95% interval sits below zero.
02

The week, every Monday

From 08:00 UTC every Monday, each brand's summary compares the last seven days with the seven before, names the weakest engine, and lists new gaps and searches you don't track.

  • A change tested on the same prompts and engines
  • The weakest engine of the week
  • Links to where the work is
Monday summary
JBL, your week in AI answers Monday, 08:00 UTC
Last 7 days49%42–55%
The 7 before44%38–50%

On the prompts both weeks asked on the same engines, the change could be run-to-run noise.

  • Weakest engineGrok
  • New gaps2
  • Searches you don't track3
03

Webhooks for everything else

Up to five HTTPS endpoints per workspace receive run.completed, recommendation.created and article.published, signed the Standard Webhooks way and retried for about a day.

  • webhook-id, timestamp and HMAC signature
  • Each event delivered once per endpoint
  • Schemas in the OpenAPI document
A webhook delivery
run.completed, abridged, signed the Standard Webhooks way
POST /hooks/rookatlas HTTP/1.1
webhook-id: msg_2xKc9Qe7
webhook-timestamp: 1793340120
webhook-signature: v1,K5oZfzN95Z9UVu1EsfQmfVNQhnkZ2pj9o9NDN/H/pI4=
content-type: application/json

{
  "type": "run.completed",
  "timestamp": "2026-11-09T06:42:00.000Z",
  "data": {
    "brand_name": "JBL",
    "status": "completed",
    "trigger": "schedule",
    "engines": ["chatgpt", "gemini", "perplexity", …],
    "answers": 62,
    "credits_charged": 8,
    "visibility": { "value": 0.645, "low": 0.474, "high": 0.786, "confidence": "low" },
    …
  }
}

How it works

  1. Step 1

    Connect a channel

    An incoming webhook per brand, checked with a first message.

  2. Step 2

    Choose your alerts

    Each person picks theirs by email; Slack takes the brand's.

  3. Step 3

    Act on what moved

    Every alert links to the runs, gaps or searches behind it.

What's inside

  • Slack per brand

    Finished and failed runs, drops and the Monday summary, from Starter.

  • Email alerts

    Failed runs, drops and the summary, per person, on every plan.

  • Drop rule

    Fires only on a fall the paired test supports, on the same engines.

  • Monday summary

    The week against the week before, with new gaps.

  • Signed webhooks

    Runs, actions and articles to your own endpoints.

  • Sealed secrets

    Slack webhooks sealed at rest and limited to Slack's host.

What it can't do

The drop rule compares one run with the one before, conservatively, so a slow slide over several runs shows in the Monday summary and the trend rather than as an alert. Finished runs don't send email, so failures and drops stand out.

All features

Questions

Prices, credits and plans are on the pricing page .

Which plans include Slack alerts?

Starter and up. Email alerts and webhooks work on every plan, Free included.

Do I need to install a Slack app?

No. Paste an incoming webhook URL for the channel; RookAtlas checks it with a first message before keeping it. Each brand can post to its own channel.

How are webhooks signed?

The Standard Webhooks way: a webhook-id, a webhook-timestamp and an HMAC-SHA256 signature, so receivers can use existing libraries and drop repeats.

Pick a plan. Every engine is included.

Alerts and summaries is on every plan, Free included. One credit runs a prompt on all eight engines (Free's five monthly credits run on ChatGPT, Gemini and AI Mode), and every run shows its price before it starts. Compare the plans and work out your credits.

Free tool, usable today: GA4 AI traffic segment

Opening soon: sign-up opens with live measurement, after our provider pilot.

Plans from
$36/mo
Engines on every paid prompt run
8
Credits per prompt
1