Getting Started
docs/Getting Started/Getting Started

Getting Started

Turn any URL into a searchable knowledge base in 2 minutes.

KnowledgeSDK extracts structured knowledge from any URL — product pages, documentation sites, landing pages — and makes it instantly searchable with hybrid keyword and vector search. Feed it a URL, get back organized knowledge items you can query with natural language.

Most users only need two endpoints: /v1/extract and /v1/search. Extract knowledge once, then search it as many times as you need.

Quick Start

1
Get your API key

Create your account at knowledgesdk.com/login?signup=1 to get your API key. Your key starts with sk_ks_ and is used in the x-api-key header for all requests.

2
Extract knowledge from a URL

Send a URL to the extract endpoint. KnowledgeSDK will crawl the site, classify the business, and break the content into structured, searchable knowledge items.

The sync endpoint typically takes 1-3 minutes depending on the site size. For non-blocking usage, see the async and streaming variants.

3
Search the extracted knowledge

Once knowledge is extracted, search across it using natural language queries. Results are ranked using hybrid keyword + vector search for the best accuracy.

What happens during extraction?

When you call /v1/extract, KnowledgeSDK performs several steps automatically:

  1. Sitemap discovery — Finds all pages on the site via sitemap.xml and link crawling.
  2. Page scraping — Converts each page to clean markdown.
  3. Business classification — Identifies the company name, domain, category, and description.
  4. Knowledge extraction — Breaks content into structured knowledge items with titles, descriptions, and categories (PRODUCT, FEATURE, PRICING, FAQ, SUPPORT, COMPANY, etc.).
  5. Vector indexing — Embeds each knowledge item for semantic search.

All of this happens in a single API call. The extracted knowledge is stored and searchable immediately.

Next Steps