SearchCans

The sites use heavy JavaScript to load their rate tables, and traditional scrapers just grab empty divs

16 views

🔍 Click to enlarge

A data analyst at a fintech startup builds a competitor monitoring system that needs to pull pricing information from 50 insurance company websites every morning. The sites use heavy JavaScript to load their rate tables, and traditional scrapers just grab empty divs. SearchCans' Reader API renders these JavaScript-heavy pages in the browser, then converts the fully-loaded content into clean Markdown that feeds directly into the company's LLM for structured extraction. The analyst sets up parallel requests through both the SERP API to discover new competitor sites and the Reader API to process them simultaneously, no hourly caps blocking the workflow when she needs to scale up to 200 sites.

A developer working on a legal research chatbot connects SearchCans to LangChain so attorneys can ask questions roughly case law. The SERP API pulls Google and Bing results for relevant precedents, while the Reader API immediately converts those court opinion URLs into Markdown that the RAG system ingests. The whole loop runs in under 3 seconds for most queries. Latency hits 1.5 seconds for standard requests without JavaScript rendering. The structured JSON output from the SERP API means the developer doesn't waste time parsing HTML — case citations, snippets, and URLs arrive ready to pipe into the next stage.

Both APIs share the same credit pool. One credit works for either a search or a URL conversion. The Standard plan gets 20,000 credits for $18, which works out to $0.90 per thousand requests. That same volume costs around $180 with alternatives like SerpApi or Firecrawl. The Starter plan drops the per-request cost to $0.75 per thousand with 132,000 credits for $99. No monthly billing. Credits last six months. New users get 100 free credits at registration to test both engines before committing.

The service handles POST requests with JSON payloads and Bearer token authentication. Developers configure timeout parameters separately for server-side processing and client-side waits — one example shows 20 seconds for server timeout and 30 for client. The wait time for JavaScript rendering adjusts based on how complex the target page is.

It breaks down when you need real-time monitoring with sub-second updates. That 1.5-second floor won't work. The Standard plan caps parallel lanes at two, so teams running dozens of simultaneous workflows need to jump to higher tiers or queue requests. Credits expire after six months, which punishes sporadic usage patterns — a researcher who needs 50,000 requests twice a year ends up buying credits that disappear before the second project starts.

The LangChain and LlamaIndex integrations matter most for developers already building in those ecosystems. If you're working in a different framework, you're writing your own glue code.

A solo developer prototyping a job aggregator that scrapes listings from 30 career sites pays $18 once and runs for months. An enterprise team processing 500,000 URLs monthly finds the credit math less appealing than flat-rate plans with unlimited usage, even at supposedly higher prices — the mental overhead of tracking credit burn across multiple projects creates friction.

Skip this if you need historical search data beyond what Google and Bing currently return, or if you're scraping sites that actively block browser automation regardless of rendering capability. Also wrong fit for teams that need customer support faster than email response times.