Jev Web Analyzer
Fetches a public SaaS page as Markdown, then asks Jev ten bounded questions about likely first-visit comprehension; URL validation, caching, sanitization, and presentation remain…
Fetches a public SaaS page as Markdown, then asks Jev ten bounded questions about likely first-visit comprehension; URL validation, caching, sanitization, and presentation remain deterministic. [Demo](https://replynodes.com/jev-web-analyzer/).
# Jev Web Analyzer
**Learn Jev with real web data.**
Jev Web Analyzer is a small, inspectable example of using **Jev** to make typed probabilistic judgments over live website content.
Paste a URL. **ReplyNodes** fetches the page and turns it into clean Markdown. **Jev**, through **Vercel AI Gateway**, evaluates that state with structured questions and returns decisions with probabilities.
**[Try the live demo →](https://replynodes.com/jev-web-analyzer/)**
<table>
<tr>
<td align="center" width="220">
<a href="https://replynodes.com/"><img src="./public/brand/replynodes.svg" alt="ReplyNodes" height="34"></a><br>
<sub>Live web context</sub>
</td>
<td align="center" width="220">
<a href="https://typesafe.ai/"><img src="./public/brand/typesafe-jev.svg" alt="Jev by TypeSafe AI" height="34"></a><br>
<sub>Typed probabilistic judgments</sub>
</td>
<td align="center" width="220">
<a href="https://vercel.com/ai-gateway"><img src="./public/brand/vercel.svg" alt="Vercel" height="28"></a><br>
<sub>AI Gateway</sub>
</td>
</tr>
</table>
> **Unofficial community project, not affiliated with TypeSafe AI.**
## What this demo teaches
The SaaS website teardown is only an example workload. The main purpose of this repo is to make Jev's evaluation pattern easy to see, run, modify, and reuse.
| Jev capability | How this project demonstrates it |
| --- | --- |
| **Choice judgments** | Infer audience, clarity, differentiation, CTA, trust, and product motion |
| **Boolean judgments** | Add your own yes/no evaluation |
| **Score judgments** | Add your own ordered rubric |
| **Probabilities** | Inspect the distribution behind a decision |
| **Multiple judgments** | Evaluate the same state against many questions in one call |
|