Web data infrastructure for the AI era
HushSearch gives AI agents, RAG pipelines and LLM applications reliable access to the live web: 35 endpoints, one key, one response format.
Why we built this
Every serious AI application needs to search and read the web. Doing that well usually means juggling a search API, a scraper, a headless browser and a PDF converter, each with its own key, pricing and response format. We wanted one envelope for all of it.
How it works
Queries go through private SearXNG metasearch, which fuses results from many engines. Pages are fetched with fallbacks from plain HTTP to a real browser, and documents are converted with Docling. A Rust gateway handles authentication, credits and caching in front of the Python API and Go workers.
Privacy-first
We do not sell query data. Usage logs store an HMAC-SHA256 hash of each query and a coarse category, never the raw text. Credits are deducted atomically in Redis and every movement is written to a ledger.
Yours to run
The whole stack ships with Docker Compose, Kubernetes manifests and Terraform, so a team can run its own post office inside its own network with the same API.
Ready to start?
500 free credits. No credit card. No lock-in.