Hello World — Introducing HushSearch
· 4 min read · Updated
The problem
Every serious AI application needs to read the web in real time. You need to search, scrape, extract and process web content continuously, and that usually means stitching together a search API, a scraper and a document converter, each with its own key, pricing and response format.
What we built
HushSearch started with Web Search, Content Extraction, Page Scrape and Search & Summarize behind one API key, paid for with credits instead of a subscription. Credits never expire.
Privacy architecture
Your query is never stored in plaintext. Usage logs keep an HMAC-SHA256 hash of the query text and a coarse category. Credits are deducted atomically in Redis using a Lua script, with every movement recorded in a ledger.
What's next
Site Crawl, Research Agent, Batch Scrape, PDF/OCR and Browser Sessions.
Update, September 2026: all of those are now live, along with embeddings, rerank, change monitoring and an MCP server. See the docs for the full list, and pricing for current credit packs. This post originally quoted a per-search price and competitor comparisons that no longer reflect today's packs; we've removed them.