Any URL → clean, LLM-ready Markdown. JS-rendered, with crawl & map built in. JS rendering, site crawls, and semantic chunking included — so your RAG pipeline starts at the interesting part.
the web, as it actually is
<div class="nav-x9" data-tr="…">
<script>window.__cfg={...}</script>
<div style="display:none">SALE!</div>
<h1 class="ttl-x">Pricing</h1>
<td>Basic</td><td>$29/mo…what your model gets
# Pricing | Plan | Price | |-------|--------| | Basic | $29/mo | Semantic chunks: 4 Tokens: 182
One URL → Markdown. Static fetch by default, real browser rendering when the page needs it.
Point at a site, get the pages that matter — sitemap-aware, budget-capped.
Semantic, heading-aware chunks sized for embeddings. Free with every plan.
Pages roll over (capped). Top-ups: 10,000 pages for $5.
GemScrape is the paid product. The rest of the suite ships around it — as free features and add-ons, not extra subscriptions. Here's what's live and what's next.
Crawl a whole site or repo into one clean, LLM-ready bundle — paste a single file into your model's context instead of wiring up a crawler. Want to know the moment it ships?
Semantic, heading-aware Markdown chunking sized for embeddings. Already runs on every scrape — free on all plans, including free.
Keep sources fresh — re-scrape on a schedule and get a webhook only when the content actually changed.
Vision extraction for image-heavy pages — alt-text and figure descriptions in your Markdown, using your own model key (BYOK). Optional; nothing in the core pipeline depends on it.