# We score ourselves in public. Same rubric we use for you, published with the checks we still fail.

URL: https://code4x.dev/why-us

Code4X's own site, code4x.dev, scores 94 out of 100 on the same rubric we use for clients.

Where we still lose points:

- Agent commerce readiness: fails
- Agent payments: fails
- Content freshness: partial

## Where we fit. Next to the tools you already know.

| What it covers | Code4X | AI visibility tools | Content AEO | Technical SEO audit |
|---|---|---|---|---|
| Checks whether AI crawlers can reach your site | Yes | No | No | Partly |
| Checks what AI reads without JavaScript | Yes | No | No | Partly |
| Checks forms and buttons an agent can use | Yes | No | No | No |
| Evidence for every lost point | Yes | No | No | Partly |
| Engineers implement the fixes | Yes | No | No | No |
| Nothing goes live without your approval | Yes | No | No | No |
| Re-check after each fix goes live | Yes | No | No | No |
| Advice on copy and topics | Yes | No | Yes | No |
| Tracks whether AI mentions your brand | Yes | Yes | No | No |

## Every point has a receipt. Here's ours, from one check to one number.

### Every check passes or fails on evidence.
A request, a rule and a result, kept for every check. This one is on our own site: `GET code4x.dev/robots.txt` returns 200, and each AI crawler is named, with a signal for search, AI answers and training. It passes.

```
User-agent: OAI-SearchBot
User-agent: PerplexityBot
User-agent: Claude-SearchBot
Content-Signal: search=yes, ai-input=yes, ai-train=yes
Allow: /
```

### Checks roll up into layer scores.
Each layer scores 0 to 100 from its checks. Each check scores for being there, and for being good. On code4x.dev, every Discovery check passes: Discovery scores 100/100.

### Layer scores combine into one number.
Discovery counts for 30%, Readability for 45% and Interaction for 25%. Every report lists the weights. code4x.dev: Discovery 100%, Readability 95%, Interaction 85%, for 94/100.

### We monitor how AI talks about you.
Every 3 weeks we check whether AI mentions you, cites you and describes you right, and show what improved.

## Four layers, one score you can trust. Same pages, same score. Rules decide most of it.

Rules set most of your score. A few checks need judgement, like whether your main heading says what you do. One fixed AI model scores those, the same way every time.

### Discovery
Layer 1, scored. Can AI find your site and understand what it offers? We check that AI crawlers are let in, pointed to the pages that matter, and told what they may use.

- AI search and assistants reach every page that matters
- Clear, separate rules for AI training and AI answers
- New pages found faster

### Readability
Layer 2, scored. Can AI read, extract and cite your content? We check that your content is in the page, well structured, accurate and current, so AI can quote it correctly.

- Prices, products and answers reach AI without JavaScript
- The same facts about you on every page
- Pages that are easy to cite as a source

### Interaction
Layer 3, scored. Can AI agents act on your site? We check that an agent can fill in a form, search, book or buy, and isn't blocked at the door.

- Agents complete enquiries and bookings for buyers
- Forms and buttons an agent understands
- Ready for agent shopping and payments as they arrive

### Representation
Layer 4, monitored every 3 weeks. How does AI describe and recommend you over time? Every 3 weeks we track how AI engines mention, cite and describe you, and show what improved.

- Know whether AI mentions and cites you
- Catch wrong or outdated descriptions early
- See which sources AI relies on

## Questions before you choose us. Short answers, from the people who built it.

### Do AI crawlers run JavaScript?
Most don't. They read the HTML your server sends and skip the scripts. Googlebot does run JavaScript, so a page can rank on Google and still look empty to AI crawlers.

### Do I need an llms.txt file?
Yes, it helps. It's one of the first files AI tools look for, and we check what's in it, not only that it's there. It won't fix pages AI can't read, so we check both.

### Can I block AI training but allow AI search?
Yes. Training crawlers and search crawlers have separate names, so robots.txt can treat them differently. You can disallow GPTBot, which collects training data for OpenAI, and allow OAI-SearchBot, which fetches pages for ChatGPT search.

### Is agent readiness the same as AEO or GEO?
They're closely linked. AEO and GEO are names for making your site easy for AI to find, read and cite. Agent readiness is the part we measure on your own site: whether agents can reach your pages and use them.

### Will the same site always get the same score?
Yes, for the same pages and the same rubric version. The AI model behind a few checks is fixed too, and if we ever change it, the report says so.

### Does a language model decide my score?
Rules decide most of it. A few checks need judgement, like whether your main heading says what you do. Those use one fixed AI model, run the same way every time, and each one is tested against pages our engineers checked by hand before it counts.

## See the difference on your own site. Then let our engineers take it from there.

- Free check: [Am I AI-ready?](https://code4x.dev/am-i-ai-ready)
- [How we scored our own site](https://code4x.dev/blogs/we-audited-our-own-website)