Make every page crawlable by every AI search engine.

The problem in 30 seconds

Modern web frameworks — Next.js, Vite + React, Vue 3, Angular — ship a minimal HTML shell and build the page on the user’s device. That’s fine for humans (their browser runs the JavaScript) but a disaster for AI bots:

• GPTBot, ClaudeBot, PerplexityBot, Bingbot do not execute JavaScript at all
• Googlebot does, but on a delayed second pass that often misses recent changes
• Your site looks like an empty <div id="root"> to every bot that matters
• You disappear from AI search citations and lose organic discovery

What an AI crawler sees on a CSR site: a 500-byte empty shell. There is nothing to quote, cite, or index. Compare to the prerendered version on the right: every link, heading, structured-data block already in the HTML.

How it works

One small Worker on your Cloudflare zone intercepts every request, asks our engine for the prerendered HTML, and serves it. The engine handles Browser Rendering, caching, and content transforms.

Request lifecycle

The edge worker treats bots and humans differently. Bots can wait for a sync render because they don’t mind a 10-second response — partial HTML is worse than delayed HTML. Humans never wait: they get either a cache HIT (instant) or origin fallback while the cache warms in the background.

Real-world numbers

We tested against www.cdvalet.com (Next.js, no SSR, WAF blocks Googlebot UA at the edge) and sportsbook.fanduel.com/navigation/nba (PerimeterX-protected SPA).

What you get

• Bot detection for 30+ crawlers including GPTBot, ClaudeBot, PerplexityBot, OAI-SearchBot, Bingbot, Applebot
• HTMLRewriter content transforms per domain: inject JSON-LD, rewrite meta tags, set titles, prepend head fragments
• Server-side screenshot endpoint for demos and visual regression
• Static-preview mode (?static=1) for sites with aggressive anti-bot
• Render-log analytics in D1: cache hit rate, latency, bot mix per domain
• Stripped script output OR full-script output configurable per request
• Sub-50ms cache hits, 24h default TTL, force-fresh via query param

Try it yourself

Open these and compare:

Integrate in 3 steps

1. Register your domain. A single INSERT into our D1 table with your origin URL, wait selector, and cache TTL. Takes 30 seconds.

   INSERT INTO domains (
     host, origin_url, cache_ttl_seconds,
     wait_until, wait_selector, post_load_delay_ms
   )
   VALUES (
     'www.yoursite.com',
     'https://www.yoursite.com',
     86400,
     'networkidle2',
     'main, [role="main"]',
     2000
   );

2. Drop the edge worker on your zone. One wrangler.toml, one route pattern, one deploy. ~60 lines of code you can audit in 2 minutes.

   [vars]
   ENGINE_URL = "https://prerender.searchtides.com/render"
   
   [[routes]]
   pattern = "www.yoursite.com/*"
   zone_name = "yoursite.com"

3. Verify. Curl your URL as Googlebot and as a normal browser. Both should return fully-rendered HTML in <100ms after the first warm. View-source shows your full page content.

   curl -A "GPTBot/1.0" https://www.yoursite.com/ | wc -c
   # should be much larger than your raw SPA shell

Cost

Browser Rendering is ~$0.09 / browser-hour on Workers Paid. Typical client site (~500 unique URLs, 24h cache TTL) renders 500 pages/day &approx; 60-90 browser-minutes, or roughly $2–$3 / month. KV reads and D1 logs are negligible on top.

What to do next

If you run a JS-heavy site and want to show up in AI search, this is the missing piece. Reach out to SearchTides and we’ll have your domain rendering in under an hour.