Is your product recommended by AI Overview

Buyers ask AI which tool to use, and it answers with a shortlist. We find out whether you’re on it, who beats you, and what to change.

Check your product

Set up your dashboard in five short steps: competitors, the questions to track, the engines to ask. Pay once when it’s ready to run.

Your buyers trust this answer.

They ask an AI engine. It searches the web, reads a handful of sites, and answers with a shortlist. Type your website above and see where you would stand.

  • It reads review sites, forums and comparisons.
  • It names four or five products, rarely more.
  • The next buyer may get a different list.

An illustration, not a measurement. Your report asks the real engines, many times each.

Distribution,
ready for liftoff.

Prepare content for your channels. Review every draft, then publish through a connected destination or export it yourself.

Set up my dashboard

Publishing depends on the channel and connected account. GitHub creates a pull request; Reddit replies are prepared for manual posting.

One report. Where you stand, who wins instead, what to fix.

One page you read in ten minutes and act on the same week. Every answer is included, so you can check our reading yourself.

Your product

yourproduct.com · meeting scheduling

Example report

Engine by engine

32%

of answers named you · 126 of 400

  • Perplexity46/7561%
  • ChatGPT33/7544%
  • GeminiGemini17/7523%
  • Claude13/7517%
  • Grok8/7511%
  • AI OverviewAI Overview9/2536%

Who gets recommended instead

  1. 1Calendly78%
  2. 2Doodle51%
  3. 3Chili Piper34%
  4. 4Your product32%
  5. 5SavvyCal19%

Share of answers that named each product, across every engine and run.

Where the engines look

What to fix

Ordered by impact per unit of effort.

  1. 01Get listed on capterra.comCited 6 times in answers that skipped you. No page there names you.
  2. 02Answer the three reddit threads the engines quotereddit.com is the second most cited source in your category.
  3. 03Publish a head-to-head page against CalendlyNamed in 78% of answers, more than anyone else.

From your domain to a fix list.

You fill in three fields. We do the asking, the counting and the reading, and the report shows its working.

Read the methodology →
  1. 01

    Tell us your product

    Your domain, one line on what it does, and the category buyers search for. Nothing else.

  2. 02

    We ask the engines what buyers ask

    Questions written the way buyers type them, with your name never in them. Each one runs several times (Google AI Overview once), with live web search on.

  3. 03

    Read the answers verbatim

    Every answer is in the report, word for word, so you can check our reading against the source.

  4. 04

    Fix the gaps

    A fix list built from the sources the engines read and who holds the answer, ordered by impact per unit of effort.

See who AI recommends in your category.

What ChatGPT, Perplexity and Gemini answer when a buyer asks. One question per category, asked 3 times on each engine with live search on. Free to read, dated, with every answer quoted.

All AI rankings →

Pay once, or keep watching.Pay once for a report. Rechecks are optional.

One report for one product, with the answers, the cited sources and a prioritised fix list. Or Track: the same questions on a schedule, every run on one dashboard.One report for one product. Choose the coverage you need and get the answers, the cited sources and a prioritised fix list. No subscription required.

Quick

A straight answer on the engines that matter most.

$19once
  • Perplexity
  • ChatGPT
  • GeminiGemini
questions
15
runs each
2
answers
90
  • Badge if you qualify
  • Competitors who beat you
  • Where engines look, and whether you're there
  • Shareable report link
Get Quick

Full

Most popular

Every engine, enough runs to trust the number.

$39once
  • ChatGPT
  • Claude
  • Perplexity
  • GeminiGemini
  • Grok
  • AI OverviewAI Overview
questions
25
runs each
3
answers
400
  • Competitor share-of-voice
  • Source map: where engines look, whether you're there, how to get on
  • Questions you can win
  • Prioritised fix list
  • Badge if you qualify
  • Shareable report link
Get Full

Deep

For a crowded category where the detail decides it.

$59once
  • ChatGPT
  • Claude
  • Perplexity
  • GeminiGemini
  • Grok
  • AI OverviewAI Overview
questions
40
runs each
3
answers
640
  • Everything in Full
  • Per-engine competitor breakdown
  • Widest question coverage
  • Badge if you qualify
  • Shareable report link
Get Deep

No engine is an add-on. Each plan runs every engine shown on its card. No seats, no subscription.

Quick

A straight answer on the engines that matter most.

$19once
  • Perplexity
  • ChatGPT
  • GeminiGemini
questions
15
runs each
2
answers
90
  • Badge if you qualify
  • Competitors who beat you
  • Where engines look, and whether you're there
  • Shareable report link
Get Quick

Full

Most popular

Every engine, enough runs to trust the number.

$39once
  • ChatGPT
  • Claude
  • Perplexity
  • GeminiGemini
  • Grok
  • AI OverviewAI Overview
questions
25
runs each
3
answers
400
  • Competitor share-of-voice
  • Source map: where engines look, whether you're there, how to get on
  • Questions you can win
  • Prioritised fix list
  • Badge if you qualify
  • Shareable report link
Get Full

Deep

For a crowded category where the detail decides it.

$59once
  • ChatGPT
  • Claude
  • Perplexity
  • GeminiGemini
  • Grok
  • AI OverviewAI Overview
questions
40
runs each
3
answers
640
  • Everything in Full
  • Per-engine competitor breakdown
  • Widest question coverage
  • Badge if you qualify
  • Shareable report link
Get Deep

No engine is an add-on. Each plan runs every engine shown on its card. No seats, no subscription.

For agencies

Running reports for clients? Buy Full reports in packs.

  • Agency 5 · $149, $30 a report
  • Agency 20 · $499, $25 a report

The same Full report, run for client brands from one account, with an optional “Prepared for” line carrying your agency’s name.

Agency packs →

For agents

Run it from your agent, on your own keys, for $29 a month.

  • 3 brands a billing period
  • 100 reports a billing period
  • 10 reports in any 24 hours
  • 2 reports at once

Reports through the API, the MCP server, the CLI or the dashboard, with Track on your own keys included. Bring the keys you have: an engine without one shows as skipped.

Agent plan →

Bring your agent.

Add recomention to Claude Code or Codex, ask “is my product recommended for meeting scheduling?”, and your agent starts the report, waits, and reads it as JSON.

Our engines, or your own keys with the Agent plan.

claude mcp add recomention --env RECOMENTION_API_KEY=YOUR_API_KEY --env RECOMENTION_BASE_URL=http://localhost:3000 -- npx -y @recomention/mcp

Replace YOUR_API_KEY with a key from your account (Developers).

Bring your agent.

Add recomention to Claude Code or Codex, ask “is my product recommended for meeting scheduling?”, and your agent starts the report, waits, and reads it as JSON.

Our engines, or your own keys with the Agent plan.

curl http://localhost:3000/api/v1/rankings

Questions before you buy.

Something else? Ask us.

Isn't this just SEO?
Traditional SEO measures where you rank on a results page. This measures whether you appear inside an answer. Different surface, different inputs, different fix list.
Why doesn't this match what I see when I ask ChatGPT myself?
The engines are non-deterministic: ask the same question twice and you can get two different shortlists, and your own account’s answers are shaped by your history and settings. We ask each question several times (Google AI Overview once), with web search on and no memory of past chats, and report the share of answers that named you rather than a single reply. Every answer is included verbatim, so you can see the spread for yourself.
Can I track changes each month?
Yes. Start with a one-time report, then optionally add monthly rechecks from your account. Your first recheck and monthly charge are 30 days later. Each paid monthly invoice funds one new report, with a trend page and an email explaining what changed. Cancel future renewals from your account at any time.
What if I'm invisible?
Then you learned the expensive thing cheaply. A zero comes with the most actionable version of the report, not the least. The fix list is built from which sources the engines read and who occupies the answers.
How accurate is it?
Accurate about what the engines said, on that date, to those questions. Every answer is included verbatim so you can check the reading yourself. If we got one wrong, we correct it.
Is the badge just a sticker?
It’s issued only if you actually qualify, it’s date-stamped, and it links to a public page showing the counts behind it. If you don’t qualify you don’t get one. That’s what makes it worth having.
Do I need an account?
No. Pay, get a private link, done.

Find out if AI recommends your product.

One report, paid once, with every answer quoted. From $19.