The biggest leak: Courierly was cited in 56% of 600 answers; you in 21%.
21%
of 600 AI answers cited Relaykit. That's your Share of Agent.
n=600 · 95% range 17–24%
No earlier week yet: this is the first week measured.
Share of Agent
21%
Your true share very likely sits between 17–24% (95% range, n=600).
You win 4 of 30 buyer questions; Courierly wins 20 of 30. A question counts as won when assistants consistently cite you for it.
Share of the same answers that cited each name. Whiskers are 95% ranges.
n=600 answers · 5 of 5 assistants
Based on AI assistants' answers, which can be wrong. Check the cited sources.
Agents that tried your API or MCP server got what they needed 75% of the time.
n=32 · 95% 58–87%
Distinct agents at each step this week, and how many fell away between steps.
AI-bot requests, checked against each vendor's published ranges
Noise we removed60 removed, never counted
Collector quiet
Your MCP SDK collector has not reported for 31 hours, so the Used numbers below may be out of date. Check that it is still deployed and that its ingest key is still valid.
Check the install guideThree fixes this week, most important first.
Fix 1: Unblock OAI-SearchBot
OAI-SearchBot is how OpenAI's assistant finds and cites pages. While it is blocked you are invisible there.
Removes the "Disallow: /" that blocks OAI-SearchBot from your whole site. They then follow the same path rules as every other bot (your "User-agent: *" group), so paths you already disallow stay disallowed. Nothing else changes: training bots are left exactly as you had them (GPTBot stays allowed, ClaudeBot stays allowed, CCBot stays allowed, Google-Extended stays allowed).
Download fix 1 as a .patchWe re-check on Oct 26 and tell you whether it worked.
Fix 2: Make github.com/relaykit/relaykit-node worth citing
Fix 3: Make relaykit.example worth citing
15 plays from our playbook apply to Relaykit.
Each one is cited, dated, and matched to your own data. A person approves every entry before it reaches you.
Sample report: the details and steps open inside your own workspace.
They apply to you, and we can see why in your own data.
Sources:OpenAI (developers.openai.com)Anthropic (support.claude.com)Perplexity (docs.perplexity.ai)
Sources:Anthropic (support.claude.com)
Sources:OpenAI (developers.openai.com)
Sources:Perplexity (docs.perplexity.ai)
Sources:Primetra (primetra.ai)
Sources:Primetra (primetra.ai)
Sources:Primetra (primetra.ai)Cloudflare (developers.cloudflare.com)
Your go-to-market plan for the next four weeks.
Buyer questions where someone else gets cited and you mostly don't.
“How do I choose an email API for a product that sends 100,000 emails a month?”
You: 1 of 20 answers · Courierly leads 70%
Source fact: found.question.33333333-0000-4000-8000-000000000011.share“Compare transactional email APIs by pricing and deliverability”
You: 2 of 20 answers · Courierly leads 80%
Source fact: found.question.33333333-0000-4000-8000-000000000006.share“Which email API has the most reliable webhooks for delivery events?”
You: 1 of 20 answers · Courierly leads 75%
Source fact: found.question.33333333-0000-4000-8000-000000000008.shareThird-party pages the assistants already cite on your questions. Earn a mention there.
Cited 179 times in answers that skipped you, by Claude, ChatGPT, Gemini, Grok, Perplexity.
Example page: reddit.com/
Cited 125 times in answers that skipped you, by Claude, ChatGPT, Gemini, Grok, Perplexity.
Example page: emailtools.example/
Cited 108 times in answers that skipped you, by Claude, ChatGPT, Gemini, Grok, Perplexity.
Example page: devtoolsdigest.example/
3 done · 4 to do · 1 not checked. The ways agents read and use a product, checked against your site.
Docs as markdown
To doServe every docs page as markdown (the same URL plus .md, or on Accept: text/markdown) so agents read it without the page chrome.
Source fact: gtm.channel.markdown_pagesOpenAPI file
To doPublish your OpenAPI file at /openapi.json and link it from the docs home so coding agents read the whole API in one fetch.
Source fact: gtm.channel.openapiMCP registry listing
To doAssistants cite an MCP registry on answers that skip you. Check whether you are listed there, and submit your server if not.
Source fact: gtm.channel.mcp_registry_listingAI search bots allowed
To doStop refusing OAI-SearchBot: allow them in robots.txt and in your CDN or firewall rules.
Source fact: gtm.channel.robots_ai_searchDocs with code examples
Not checkedGive every docs page a copy-paste example that runs as written, with the language and the exact endpoint.
llms.txt
DoneMCP server
DonePricing readable as text
DoneWeek 1Open the doors
Week 2Answer the questions
Week 3Earn the mentions
Week 4Measure and repeat
Built from 600 answers across 5 assistants this week; sources are what the assistants cited on answers that skipped you, not a ranking of the web. · method found-v2 · used-v4 · fix-v1
Latest agent activity8 events · last at Oct 18 · 23:10 UTC
ClaudeBot fetched /ok (200)
ok (200)OAI-SearchBot fetched /changelogok (200)
ok (200)Perplexity-User fetched /pricingok (200)
ok (200)OAI-SearchBot fetched /docs/webhooksblocked (403)
blocked (403)ChatGPT-User fetched /ok (200)
ok (200)PerplexityBot fetched /blog/:slugok (200)
ok (200)ChatGPT-User fetched /docs/:slugok (200)
ok (200)chrome fetched /docs/quickstartok (200)
ok (200)