Loading
Is your business ready for AI agents?
Get a readiness score out of 100 that shows where AI agents get stuck on your website, your sign in and your phone line. The Agent Readiness Index is free and comes from Encore AI. 1,359 organizations are scored so far, with an average of 42.6 out of 100.
To scan a website, open this page in a browser and enter its address. The scan takes about two minutes and reads public pages only. Agents can read every report as markdown or JSON without scanning anything.
Six areas, 100 points
| Area | Points | The question it answers |
|---|---|---|
| Discoverability | 15 | Can an agent find and read your pages? |
| Protocol access | 20 | Can an agent connect to you through MCP or an API? |
| Identity and delegation | 20 | Can a customer give an agent limited access without sharing a password? |
| Web experience | 15 | Does your site work when an agent uses it? |
| Voice and IVR | 20 | What happens when an agent calls your phone line? |
| Policy and governance | 10 | Do your terms allow agents that act for customers? |
Six levels of readiness
- Level 0, Closed. Declared agents are blocked from the public website.
- Level 1, Tolerated. Pages load for an agent. Nothing is built for one.
- Level 2, Legible. An agent can read and cite your content.
- Level 3, Connectable. An agent can connect through an MCP server or a public API program.
- Level 4, Trusted. A customer can authorize an agent with limited, revocable access.
- Level 5, Native. Agents are a supported channel on web, API and voice.
Where organizations stand today
Across 1,359 scored organizations the average is 42.6 out of 100 and the middle score is 39.0. 214 score 60 or more and 48 score below 20. The highest score is 98 and the lowest is 4.
| Level | Organizations | Share |
|---|---|---|
| Level 0, Closed | 321 | 24 percent |
| Level 1, Tolerated | 136 | 10 percent |
| Level 2, Legible | 806 | 59 percent |
| Level 3, Connectable | 67 | 5 percent |
| Level 4, Trusted | 29 | 2 percent |
| Level 5, Native | 0 | 0 percent |
Checks most organizations fail
- Poppy file for personal agents: 100 percent fail, 0 percent pass.
- Agent card for agent to agent calls: 100 percent fail, 0 percent pass.
- Browser tools exposed to agents: 99 percent fail, 1 percent pass.
- MCP server: 90 percent fail, 7 percent pass.
- Developer and data sharing program: 86 percent fail, 0 percent pass.
The leader in each industry
| Industry | Leader | Score |
|---|---|---|
| Banks | M&T Bank | 65 out of 100 |
| Lenders | Prosper | 70 out of 100 |
| Insurance | Lemonade | 63 out of 100 |
| Payments and fintech | Brex | 91 out of 100 |
| Telecom and internet | Deutsche Telekom | 75 out of 100 |
| Utilities | Southern California Edison | 60 out of 100 |
| Airlines | JetBlue | 63 out of 100 |
| Healthcare | Cigna | 66 out of 100 |
| Retail | Amazon Web Services, Inc. | 76 out of 100 |
| Media and streaming | Spotify | 67 out of 100 |
Recently scored
- fnac.com, 4 out of 100
- Vinted, 33 out of 100
- maidenform.com, 4 out of 100
- Gildan, 31 out of 100
- Bali.com, 39 out of 100
- Horchow, 20 out of 100
- Lastcall, 4 out of 100
- OfferUp, 42 out of 100
- Moncler France, 30 out of 100
- Louisvuitton, 23 out of 100
- Iberia, 48 out of 100
- Grailed, 35 out of 100
- GOAT, 36 out of 100
- Burberry, 54 out of 100
- Budget | Rent a Car, 43 out of 100
- LLMagnet, 70 out of 100
- Bose, 39 out of 100
- Blizzard Entertainment, 36 out of 100
- Asics, 31 out of 100
- Arc'teryx Equipment, 53 out of 100
- Ancestry, 20 out of 100
- AMC, 42 out of 100
- Allegiant, 20 out of 100
- Alamo, 35 out of 100
Frequently asked questions
- What is the Agent Readiness Index?
- The Agent Readiness Index is a free, public index by Encore AI. Enter an organization's web address and it scores, on a scale of 0 to 100, how ready that organization is for personal AI agents that act for customers. Every report lists the checks, the evidence and a fix for each gap.
- What does AI agent readiness mean?
- It means how well an organization works when a customer's AI agent, such as ChatGPT, Claude, Gemini or Perplexity, tries to do a job for them: find an answer, check an account, change a booking or call the phone line. A ready organization lets a declared agent read its pages, offers a documented way to connect such as an MCP server, and lets a customer approve limited, revocable access without sharing a password.
- How is the readiness score calculated?
- The score adds up six areas worth 100 points: Discoverability 15, Protocol access 20, Identity and delegation 20, Web experience 15, Voice and IVR 20, and Policy and governance 10. The five website areas come from published checks and are calibrated with one public curve. The voice area comes from recorded calls to the organization's published phone number.
- What do the readiness levels mean?
- There are six levels. Level 0, Closed: declared agents are blocked. Level 1, Tolerated: pages load, nothing is built for an agent. Level 2, Legible: an agent can read and cite the content. Level 3, Connectable: an agent can connect through an MCP server or a public API program. Level 4, Trusted: a customer can authorize an agent with limited, revocable access. Level 5, Native: agents are a supported channel on web, API and voice.
- Does a scan cost anything, and how long does it take?
- A scan is free. The website checks take about two minutes. The phone line tests run separately and are added to the report when they finish.
- What does the scan read, and is it safe for my site?
- The scan reads public pages and published technical files only, such as robots.txt, llms.txt, a sitemap and well known metadata. It identifies itself as an agent with a user agent that names Encore AI. It loads the sign in page to see how an agent is treated, but it never types into a field and never tries to sign in.
- Which organizations are scored?
- 1,359 organizations are scored so far, across 25 kinds of business, among them large banks, insurers, airlines, retailers, telecom companies, health systems, software companies and restaurants. Anyone can scan another public website, and the new report is added to the index.
- How can an organization raise its score?
- Every report lists each gap with the evidence the scan found and a fix. The largest single check is a live MCP server, worth 8 points. Cheaper steps include publishing llms.txt, adding structured data, rendering content on the server, publishing a sitemap, allowing declared agents in robots.txt, and stating a clear position on AI agents in the terms.
More
- Leaderboards: the top five in ten industries
- Method: every area, check, level and call test
- Research: how all scored organizations do
- Articles: written from the findings in the reports
- For AI agents: markdown, JSON, an MCP server and our policy