Skip to main content
ai-cmo.dev

Buyers ask ChatGPT. It names someone else.

The GEO agent asks the questions your buyers ask on 7 answer engines, shows which answers skip you, and rewrites the pages, llms.txt and schema until the next answer names you.

  • Free forever
  • No card
  • First score in about 30 seconds
Works with
  • Claude Code
  • Cursor
  • Codex
  • Windsurf
  • Devin
  • Copilot

The difference

More than another report

demo dataai-cmo.dev
72Visibility score
SEO
72
GEO
58
Where you show up
  • Perplexity44%
  • Gemini60%
Conversions · 30d
Visits12,800
Signups1,462

7 answer engines · prompts re-asked every run · llms.txt + JSON-LD

The problem

You cannot see the answers that are losing you deals.

Buyers ask ChatGPT what to use. It answers. You are not in it. No dashboard you own shows that conversation.

  1. 01right now

    Buyers ask ChatGPT, not Google, and you cannot see those answers.

  2. 02right now

    Competitors get cited for questions your product answers better.

  3. 03right now

    llms.txt and structured data rot quietly. Nothing alerts you.

The GEO agent

One agent reads the answers. Then it fixes why you are not in them.

Four steps. Every run. Nothing ships without you.

  1. 01

    Ask

    Your tracked buyer prompts run on 7 answer engines.

  2. 02

    Compare

    Citations are scored against your competitors, per engine.

  3. 03

    Draft

    llms.txt rewrites, JSON-LD blocks and entity-page fixes land as drafts.

  4. 04

    Re-measure

    The next run shows whether the engines picked them up.

Plays well with others

The GEO agent does not work alone. It hands off to the other nine. Same queue, same score.

What it costs

Free. Like, actually free.

The GEO agent is available on Free. Free includes 5 one-time credits, with no card. We meter work, not seats.

Free

$0/mo

Know exactly what Google and the LLMs say about you. Fix it yourself.

5 one-time credits

  • GEO agent
  • SEO agent
  • X agent

The one-time grant never expires · No team seats · 1 website

Start free

Pro

$249/mo

All ten agents, Google Ads and technical docs included.

2,000 credits each month

  • GEO agent
  • SEO agent
  • Research agent
  • Articles agent
  • Docs agent
  • Reddit agent
  • X agent
  • Hacker News agent
  • LinkedIn agent
  • Google Ads agent

Monthly plan credits do not roll over · 10 team seats · 1 website

Choose Pro

ships on this plan measures only Compare all plans

Built for developers

One prompt. Your coding agent does the rest.

Copy it, paste it into your agent. It installs ai-cmo, pulls your score and ships the fixes as a PR you review.

prompt.txt
Install the ai-cmo MCP server (npx -y ai-cmo-mcp@0.1.0), score ai-cmo.dev, and fix everything in my review queue. This one-liner is the quick start. The full guided setup for ai-cmo.dev begins with https://ai-cmo.dev/llms.txt and https://ai-cmo.dev/openapi.json, resolves the workspace with aicmo_whoami({}), and ends at the install page https://ai-cmo.dev/api/v1/workspaces/$WORKSPACE/install. Read https://ai-cmo.dev/llms.txt and https://ai-cmo.dev/openapi.json. If npx returns 404, report the unavailable package and stop. Run aicmo_whoami({}); expect authorized workspace access. For each candidate, set WORKSPACE to its returned id and read authenticated GET https://ai-cmo.dev/api/v1/workspaces/$WORKSPACE/settings with Authorization: Bearer $AICMO_API_KEY (settings:read), or use verified settings already supplied by the client. Expect domains to contain ai-cmo.dev; stop if no matching workspace or authorized settings are available. Run aicmo_score and aicmo_recs_list with the matching workspace id. Follow each returned recommendation’s brief and verification steps, run the repository checks and expect them to pass. Open a pull request for review; do not publish, deploy or merge.
Claude CodeCursorCodex

Stop being the product ChatGPT forgets.

  • First score in about 30 seconds
  • No card, no sales call
  • Nothing ships without your approval

Fine, ask

Things you will ask before turning it on.

All 7 of them: ChatGPT, Claude, Gemini, Perplexity, Grok, DeepSeek and Llama. Five answer with live retrieval and two from model memory. Each run asks the same stored buyer questions on every engine and records whether you are mentioned, where, and which sources get cited.
Your domain and an email address. The agent reads your pages and writes the buyer questions itself, so there is no connector to set up before the first run. Every question stays editable in settings: rewrite it, archive it or add your own.
One item per gap, with the prompts it lost, a verbatim excerpt of the answer that skipped you and the fix: an llms.txt rewrite, a JSON-LD block or an entity-page change. A gap only becomes an item when its evidence carries at least two prompt ids and one excerpt.
A 0 to 100 number with five weighted parts: share of voice 35%, visibility 25%, average position 15%, modeled reach 15%, sentiment 10%. A part that cannot be calculated is dropped and the rest re-normalized, so a missing signal never counts as zero.
Monthly on Free, weekly on Lite, daily on Pro, or on demand from the dashboard. Every run re-asks the same question set, so a real change is distinguishable from a re-roll. Switching the agent off changes nothing about the others.
Free includes the GEO agent with 5 one-time credits and no card, enough for the first score. A full measurement on the standard preset reserves about 315 credits for a reference brand; your own hold is the per-brand estimate shown before each run, and it settles at the recorded provider cost, never more, with one credit being $0.20 of that cost plus a fixed surplus.
No. The GEO agent is on Free with 5 one-time credits that never expire. Lite at $129 a month and Pro at $249 add 2,000 credits a month, so measurement keeps running past the free grant, and they add the research, writing and community agents.
No. Each gap lands as a proposal you accept, decline or archive, and your own coding agent applies the change and opens the PR. An item only completes once a review records the evidence.
The next run re-asks the same prompts and shows the new answers beside the previous ones, per engine. The score moves only when the engines actually pick up the change, and every response is kept so you can read what changed.

The other nine

Nine more. One per channel. Same queue, same score.

  • Free

    SEO

    Crawls your site the way Google and Bing do: canonicals, sitemap, metadata, render blocking. The day something slips, it drafts the technical fix as a PR.

    Setup required Connect your site

    See what it drafts
  • Lite

    Research

    Crawls your competitors and the questions buyers actually ask, then hands the other agents a brief on what to write and where you are losing.

    Setup required Connect your site

    See what it drafts
  • Lite

    Articles

    Takes a prompt you keep losing, writes the long-form page that wins it and publishes it on your site through the API: your blog or your docs.

    Setup required Connect your site

    See what it drafts
  • Pro

    Docs

    Audits your docs the way an engine reads them and drafts the per-tool pages ChatGPT and Claude cite when someone asks how.

    Setup required Connect your site

    See what it drafts
  • Lite

    Reddit

    Scans your subreddits for high-intent threads, checks each community's rules, and drafts the reply that helps first.

    Setup required Connect your site

    See what it drafts
  • Free

    X

    Turns shipped work, run results and your changelog into posts and threads that sound like you, not a marketer.

    Setup required Connect your site

    See what it drafts
  • Lite

    Hacker News

    Watches the front page and new threads for topics you belong in, then drafts the Show HN post and the evidence-first comment. Never a pitch.

    Setup required Connect your site

    See what it drafts
  • Lite

    LinkedIn

    Writes founder and company posts from the same evidence, tuned for how the feed actually ranks and reads.

    Setup required Connect your site

    See what it drafts
  • Pro

    Google Ads

    Turns your own spend and conversion data into daily budget proposals. You approve each one, you set the caps, and nothing in the account changes until you say so.

    Setup required Connect your site

    See what it drafts

You build the product.
We make sure people find it.