These companies sell AI agents. Can an agent use their website?
Every company below says agents are the future of software. So I ran an open-source agent scanner on each homepage and graded what it found: llms.txt, readable HTML, markdown, API docs, MCP, robots rules, and login walls.
I graded my own site first. andyrosic.com scored 53, an F. So I added an llms.txt, markdown pages, a read-only API, and an MCP server, and the rescan scored 100, an A. Everyone in this class gets the same red pen, including me. See my before and after.
The class, ranked
Homepages scanned October 30, 2026. Click any company for its full report card.
- 1AZapier 98/100Nothing to fix. Gold star.Agent builders
- 2ACrewAI 97/100Nothing to fix. Gold star.Agent builders
- 3AIntercom 94/100Nothing to fix. Gold star.Enterprise agent platforms
- 4AFactory 93/100Mentions MCP, never shows up with it.Coding agents
- 5ABrowserbase 90/100Mentions MCP, never shows up with it.Agent builders
- 6BArtisan 89/100Couldn't find the API docs. Neither could the robot.Sales and vertical agents
- 7BRetool 84/100Couldn't find the API docs. Neither could the robot.Agent builders
- 8CZendesk 79/100No MCP server in sight. Ironic, for an agent company.Enterprise agent platforms
- 9CNotion 75/100Hands in every page as HTML soup. Markdown was on the syllabus.Enterprise agent platforms
- 10CAtlassian 73/100Hands in every page as HTML soup. Markdown was on the syllabus.Enterprise agent platforms
- 11CRelevance AI 71/100Hands in every page as HTML soup. Markdown was on the syllabus.Agent builders
- 12CSalesforce 71/100Hands in every page as HTML soup. Markdown was on the syllabus.Enterprise agent platforms
- 13DAsana 69/100No llms.txt. Please see me after class.Enterprise agent platforms
- 14DGlean 69/100No llms.txt. Please see me after class.Enterprise agent platforms
- 15DLindy 69/100Hands in every page as HTML soup. Markdown was on the syllabus.Agent builders
- 16DReplit 69/100Hands in every page as HTML soup. Markdown was on the syllabus.Coding agents
- 17D11x 67/100Hands in every page as HTML soup. Markdown was on the syllabus.Sales and vertical agents
- 18DQualified 67/100Hands in every page as HTML soup. Markdown was on the syllabus.Sales and vertical agents
- 19DDatabricks 66/100Hands in every page as HTML soup. Markdown was on the syllabus.Enterprise agent platforms
- 20DAutomation Anywhere 65/100Hands in every page as HTML soup. Markdown was on the syllabus.Enterprise agent platforms
- 21DCohere 65/100Hands in every page as HTML soup. Markdown was on the syllabus.Labs and models
- 22DCursor 65/100Hands in every page as HTML soup. Markdown was on the syllabus.Coding agents
- 23DHubSpot 65/100Hands in every page as HTML soup. Markdown was on the syllabus.Enterprise agent platforms
- 24Dn8n 65/100Hands in every page as HTML soup. Markdown was on the syllabus.Agent builders
- 25DAI21 63/100Hands in every page as HTML soup. Markdown was on the syllabus.Labs and models
- 26DKore.ai 63/100Hands in every page as HTML soup. Markdown was on the syllabus.Customer service agents
- 27DWriter 63/100Hands in every page as HTML soup. Markdown was on the syllabus.Enterprise agent platforms
- 28DCognition 62/100Hands in every page as HTML soup. Markdown was on the syllabus.Coding agents
- 29DMistral 62/100Hands in every page as HTML soup. Markdown was on the syllabus.Labs and models
- 30DDecagon 61/100Hands in every page as HTML soup. Markdown was on the syllabus.Customer service agents
- 31FClay 59/100No llms.txt. Please see me after class.Sales and vertical agents
- 32FMoveworks 58/100No llms.txt. Please see me after class.Enterprise agent platforms
- 33FHippocratic AI 55/100Hands in every page as HTML soup. Markdown was on the syllabus.Sales and vertical agents
- 34FUiPath 55/100No llms.txt. Please see me after class.Enterprise agent platforms
- 35FBolt 52/100No llms.txt. Please see me after class.Coding agents
- 36FBox 52/100Couldn't check. The robot got a 403 at the door.Enterprise agent platforms
- 37FCrescendo 52/100No llms.txt. Please see me after class.Customer service agents
- 38FIBM 51/100No llms.txt. Please see me after class.Enterprise agent platforms
- 39FSierra 51/100No llms.txt. Please see me after class.Customer service agents
- 40FAWS 50/100No llms.txt. Please see me after class.Enterprise agent platforms
- 41FCognigy 49/100No llms.txt. Please see me after class.Customer service agents
- 42FxAI 49/100No llms.txt. Please see me after class.Labs and models
- 43FLangChain 48/100No llms.txt. Please see me after class.Agent builders
- 44FPoolside 48/100No llms.txt. Please see me after class.Coding agents
- 45FHebbia 46/100No llms.txt. Please see me after class.Sales and vertical agents
- 46FLovable 45/100Sells agents. Turned ours away at the front door.Coding agents
- 47FSnowflake 44/100No llms.txt. Please see me after class.Enterprise agent platforms
- 48FOpenAI 43/100Couldn't check. The robot got a 403 at the door.Labs and models
- 49FDust 39/100No llms.txt. Please see me after class.Agent builders
- 50FPerplexity 21/100Sells agents. Turned ours away at the front door.Labs and models
- 51FServiceNow 16/100Sells agents. Turned ours away at the front door.Enterprise agent platforms
- 52FAda 12/100Sells agents. Turned ours away at the front door.Customer service agents
Grade your own site
Your next customer might be an agent shopping on someone's behalf. Find out what it sees on your site before it decides you're too much work. One command gets you the same report card the class got.
What happens when you run it
What you get back
- A score from 0 to 100 and a letter grade for each of ten subjects.
- A plain reason behind every grade, so you know exactly what the robot saw.
- Fixes ranked by how many points each one is worth, so the first hour goes to the biggest gain.
- Markdown or JSON output with --md or --json, ready for a PR comment or a CI check.
- Your docs and pricing pages graded too, with --pages /,/docs,/pricing.
Live in Claude Code? Install the plugin with /plugin install agent-landing --marketplace andyuninvited/agent-landing and ask Claude to grade your site and fix the cheapest points first. After a launch, agent-landing score reads your own telemetry and tells you whether the launch landed with agent users or only with people.
Coming soon: the Robot Mystery Shopper
An agent tries to find pricing, read the docs, and sign up on famous sites, narrated like a nature documentary. Here we see the agent, lost in a cookie banner for the third time.