We scan your site
The analyzer reads your public pages and runs the 19 technical checks that shape search and answer-engine visibility.
See what gets checked
The AI CMO for developers
See how ChatGPT, Claude, Gemini, Perplexity, Grok, DeepSeek and Llama talk about your product, and how Google ranks it. One score. Nine agents draft the fixes. You review every change.
Free analysis5 credits, never expireNo credit card
Or install the MCP server first ↓
The engines that read your site
ChatGPT
Claude
Gemini
Perplexity
Grok
DeepSeek
Llama
Google
Bing
AI Overviews
Firecrawl
01 / Where demand shows up
Buyers ask an engine. The answer names someone.
Seven answer engines, plus your Google positions from Search Console. One crawl, one score, and the exact prompts where a competitor is named instead of you.
Demand sources
AI answer engines
- ChatGPT
- Claude
- Gemini
- Perplexity
- Grok
- DeepSeek
- Llama
Search
- Google
- Bing
- AI Overviews
Visibility score
measured · Aug 28, 2026
68.6
60.4
GEOChatGPT33
Claude67
Gemini50
Grok100
Perplexity100
What the agents fix
- GEOllms.txt + JSON-LD coverage
- SEOcanonicals, sitemap, indexability
- Researchcited buyer-question reports
- Articleslong-form drafts for prompt gaps
- Redditdisclosed thread replies
- Xpost drafts in your voice
- Hacker Newsplain Show HN launches
- LinkedInfounder-voice posts
- Google Adscampaign proposals under your daily cap
Every fix arrives as a draft you approve before it ships.
02 / The agents
Nine agents run the play. You stay the editor.
Each one drafts. Nothing publishes until you approve it.
GEO · agent
Free plan
finds the prompts you're losing
SEO · agent
Free plan
tracks your Google positions, flags the page that's slipping
Research · agent
Lite plan
digs up the questions buyers actually ask
Articles · agent
Lite plan
drafts the page that answers one of them
Reddit · agent
Lite plan
finds the threads where your tool belongs
X · agent
Free plan
drafts the post that answers a question buyers ask
Hacker News · agent
Lite plan
drafts plain Show HN launches from current product docs
LinkedIn · agent
Lite plan
turns each shipped fix into a founder-voice post
Google Ads · agent
Pro plan
proposes campaign changes from spend and conversion data, under your daily cap
Every draft lands as a reviewable item with its evidence attached.
Reddit, X, LinkedIn and Google Ads are trademarks of their respective owners. Use is nominative only; no endorsement implied.
03 / How it works
Four steps. You only touch one.
The agents crawl, score and draft. You approve. The next run shows what moved.
You get one score
Seven answer engines plus your Google positions roll into one visibility score, with every result behind the number.
Meet the agentsYou approve the fixes
Each agent turns findings into reviewable drafts with the evidence attached. This is the only step that needs you.
How review works
We prove it moved
An approved change lands as a diff. The next run records the score change against the same checks.
Connect GitHub
04 / MCP install
Connect your coding agent in one command.
Install once. Your agent reads findings, claims recommendations and returns diffs for review. The dashboard and your agent never disagree.
- One command, no tag manager, no pixel
- Read-only by default. Approval stays with you.
- Works with Claude Code, Cursor, Codex and Devin
Add the server, then ask for your score in the same session.
# Add ai-cmo to Claude Code and ask for your scoreclaude mcp add ai-cmo -- npx -y ai-cmo-mcpclaude> "Show my ai-cmo.dev score and what to fix first"Add the server, then ask for your score in the same session.
// .cursor/mcp.json (connect Cursor to ai-cmo){ "mcpServers": { "ai-cmo": { "command": "npx", "args": [ "-y", "ai-cmo-mcp" ], "env": { "AICMO_API_KEY": "<your API key>" } } }}Add the server, then ask for your score in the same session.
# Add ai-cmo to Codex and list recommendationscodex mcp add ai-cmo -- npx -y ai-cmo-mcpcodex> "List my latest recommendations"Add the server, then ask for your score in the same session.
- 1
Settings → Connections → MCP servers → Add a custom MCP - 2
npx -y ai-cmo-mcp
$ claude mcp add ai-cmo -- npx -y ai-cmo-mcp> Show my ai-cmo.dev score and what to fix firsttools/list · 29 toolsaicmo_whoami · 164 msaicmo_score {"workspace":"didit"} · 23 msaicmo_scoreboard {"workspace":"didit"} · 21 msaicmo_recs_list {"workspace":"didit"} · 45 msScore 60.4 · geo-score-v1 · 2026-08-28visibility 63.3 · position 52.4 · sentiment 61 · volume 63.3Fix first:1. Close the GEO evidence gap for tracked (low)2. Fix social_branding_consistent for didit (low)05 / What we check
19 checks. The same ones the crawlers run.
Observed value against expected value, pass or fail, on every run. These eight quietly kill most sites.
| Check | What we test | Why it matters |
|---|---|---|
| llms.txt, graded | Present at your root, then graded. The content has to hold up, not just exist. | It is the first file a grounded model reads about you. |
| Sitemap freshness | lastmod coverage across every URL we sample. | A stale sitemap is a slow fade out of the index. |
| JSON-LD that parses | Every block validated; missing required fields flagged by name. | Broken schema hides you from rich results and blurs your entity. |
| Canonicals that match | Each sampled URL against the canonical it declares. | A misconfigured canonical makes you invisible twice. |
| Indexability | robots rules, noindex flags and status codes on every sampled URL. | One stray noindex and you do not exist. |
| PageSpeed, mobile first | Lighthouse performance, SEO, best practices and accessibility on every run. | Slow mobile pages lose crawlers and buyers alike. |
| Answer presence | Grounded calls across 7 engines; every mention extracted and counted. | Answer presence is the visibility search rankings cannot show. |
| Docs quality | Your pages, read the way a crawler reads them. | Thin docs get skipped before the model names you. |
06 / Pricing
Free analysis. Paid execution.
Start with 5 credits that never expire. Lite is $129/mo. Pro adds the Google Ads agent.
Save 17%
Free
Measure your site. Three agents, no card.
$0/mo
No credit card required
3 agents included
- GEO agent (Included in Free)
- SEO agent (Included in Free)
- X agent (Included in Free)
- Research agent (Not included in Free)
- Articles agent (Not included in Free)
- Hacker News agent (Not included in Free)
- LinkedIn agent (Not included in Free)
- Reddit agent (Not included in Free)
- Google Ads agent (Not included in Free)
Lite
Most pickedEight agents for one product site.
$129/mo
Cancel anytime
8 agents included
- GEO agent (Included in Lite)
- SEO agent (Included in Lite)
- X agent (Included in Lite)
- Research agent (Included in Lite)
- Articles agent (Included in Lite)
- Hacker News agent (Included in Lite)
- LinkedIn agent (Included in Lite)
- Reddit agent (Included in Lite)
- Google Ads agent (Not included in Lite)
Pro
Adds the Google Ads agent.
$249/mo
Cancel anytime
9 agents included
- GEO agent (Included in Pro)
- SEO agent (Included in Pro)
- X agent (Included in Pro)
- Research agent (Included in Pro)
- Articles agent (Included in Pro)
- Hacker News agent (Included in Pro)
- LinkedIn agent (Included in Pro)
- Reddit agent (Included in Pro)
- Google Ads agentPro (Included in Pro)
1 website per plan · add more at 10% to 40% off
What lands in your queue
Hire the team. Here's the daily work contract.
What each agent delivers on your plan, every scheduled run. Runs monthly on Free, weekly on Lite, daily on Pro.
| Deliverable | Free | Lite | Pro |
|---|---|---|---|
| Answer-engine score and citation gaps across 7 engines | ✓ Included | ✓ Included | ✓ Included |
| Google position tracking, 19-check audit, ranked fixes | ✓ Included | ✓ Included | ✓ Included |
| Cited five-part market report | Not included | ✓ Included | ✓ Included |
| Long-form article drafts, rubric-checked | Not included | ✓ Included | ✓ Included |
| Ranked threads with disclosed reply drafts | Not included | ✓ Included | ✓ Included |
| 280-character answer drafts in your voice | ✓ Included | ✓ Included | ✓ Included |
| Two to three Show HN variants from your docs | Not included | ✓ Included | ✓ Included |
| One founder-voice post per shipped fix | Not included | ✓ Included | ✓ Included |
| Budget and keyword proposals under your daily cap | Not included | Not included | ✓ Included |
Review everything. Nothing publishes itself.
Old way / new way
$13,000+/mo → From $129/mo
The rows below use the same pricing data as the plan cards. Compare the monthly cost of hiring it out with what the agents include.
Hiring-it-out rates are estimates for one product site, not from a run.
| The work | Hiring it out | ai-cmo.dev |
|---|---|---|
| Marketing hire | $5,000/mo | Included |
| SEO agency | $4,000/mo | Included |
| Content writer | $1,500/mo | Included |
| Social manager | $1,500/mo | Included |
| Community management | $1,000/mo | Included |
| Answer-engine monitoring | Not possible | 7 engines, every run |
| Always-on coverage | Not possible | Scheduled runs, no shifts |
| Engines covered | Google only, checked by hand | 7 answer engines + Google positions |
| Time to fix | Weeks of handoffs | Same-day reviewed diff |
| Score ceiling | Capped: no engine signal | Full score on every engine |
| Total | $13,000+/mo | From $129/mo |
07 / Questions
The stuff you'd ask before installing.
Short answers. The docs have the long ones.
Both, weighted equally. The SEO agent tracks your Google positions through Search Console and runs 19 technical checks. The GEO agent asks ChatGPT, Claude, Gemini, Perplexity, Grok, DeepSeek and Llama what they recommend. Same run, same score, same evidence trail.
Yes, and it looks like code review. You install one MCP server, ask your agent for the queue, and get back ranked drafts and diffs against files you already own: docs pages, JSON-LD, llms.txt, sitemap config. No campaigns, no calendar, no strategy deck.
Mostly by being legible. Models cite pages crawlers can read: clear task-shaped docs titles, valid structured data, an llms.txt that covers your surface, honest comparison pages for "X vs Y" questions, fresh sitemaps, fast mobile pages. The agents flag which of those costs you the most answers, draft the fix, and re-measure after you ship.
Free includes 5 one-time credits that never expire, no card required. Lite is $129/mo. Pro is $249/mo. Paid plans include 2,000 credits every month.
No. We read your public site the way a crawler does, and we touch your repo only through your own coding agent, only where you point it. One command installs the MCP server, one removes it. No tag manager, no pixel, no DNS changes, nothing in your bundle.
Only as drafts, only through your agent. The agent prepares a reviewable change; you approve, edit or reject it like any PR. Approving queues a re-measurement so the next score proves what moved. Nothing merges itself.
Visibility calls carry one message, the measurement prompt, with provider data collection denied and zero-retention routes required. A model without a ZDR route is skipped, never silently rerouted. One documented exception: grounded Claude has no ZDR route yet, so it runs deny-only and is flagged per response.
Site analysis fires the moment your workspace exists: robots.txt, sitemap, 19 technical checks, mobile PageSpeed, structured data, llms.txt. The agents extract your product, competitors and socials next, and your first SEO and GEO scores land the same day. No sales call, no "book a demo".
ai-cmo.dev
Hire the agents. Keep the keys.
Start with the free analysis, watch the first drafts land, then let your own agent pull the queue over MCP. No card.