Shipping the Research API: lessons from a year of multi-step retrieval
How we went from prototype to production with a multi-step research pipeline that handles 10M queries/day.
— Sarah Chen, Staff Engineer
Search, Contents, and Research APIs designed to ground your AI in the open web. Docs, SDKs, a monthly challenge, and the showcase — all in one place.
The platform
Pick the right primitive for the job — or compose all three.
Sub-second web results with snippets, links, and metadata. Built for RAG.
await you.search({
query: "latest react 19 features",
freshness: "week",
});QuickstartClean, parsed page contents from any URL. Markdown, structured, ready to embed.
await you.contents({
urls: ["https://example.com/post"],
format: "markdown",
});QuickstartAutonomous research pipeline with streaming intermediate steps and citations.
for await (const step of you.research({
prompt: "Compare RAG vs fine-tuning",
})) console.log(step);QuickstartSDKs
WebFinds · Live demo
Describe what you're looking for in plain English. WebFinds extracts a checklist of criteria from your query, searches the live web, verifies each result against that checklist, and pulls out the fields that matter — into a spreadsheet you can save, share, and export.
Criteria are the binary checks each result must pass to be marked Verified. We auto-extract them from your query — edit, remove, or add your own.
Columns are extracted by the LLM from each result's title and snippet. Empty cells (n/a) mean the value couldn't be confidently determined.
Live playground
Your API key never touches a third-party log — requests proxy through this origin and are forwarded to You.com directly.
// Run a query to see the response.
// Results stream in real-time.Use the Research API to build an autonomous agent that does deep research on a topic you care about — vintage synths, pro cycling, mycology, your call. Bonus points for surprising the judges.
Prize
$5,000 cash + featured Showcase spot
Deadline
Showcase
Linear
Linear uses You.com Search to ground its AI assistant in real-time engineering documentation across the web.
Notion
Notion AI pulls fresh web context via the Contents API to answer questions beyond the workspace.
Cursor
Cursor enriches code completions with up-to-date framework docs fetched via You.com.
Perplexity Workspaces
Perplexity uses the Research API to power deep multi-step queries inside enterprise workspaces.
Replit Agent
Replit Agent calls Search API during code generation to verify library APIs.
Granola
Granola enriches meeting summaries with public context using Contents API.
Events
Integration clinics, meetups, and the occasional hack night.
Live walkthrough of building production-grade Search API integrations. Bring your code, leave with answers.
In-person meetup with the You.com team and Bay Area developers building search-powered products.
Drop-in session to ask the engineering team anything about the Research API.
4-hour virtual hack night. Prizes for the most creative API integration.
Blog
How we went from prototype to production with a multi-step research pipeline that handles 10M queries/day.
— Sarah Chen, Staff Engineer
A field guide to retrieval-augmented generation that actually works in production. Code samples in TS and Python.
— Marcus Webb, DevRel
Behind the scenes on the architectural choices that make Contents API fast, cheap, and predictable.
— Priya Anand, Eng Manager
Status
Real-time uptime and incident reports.
Changelog
Everything new across the platform.
You can now stream intermediate steps from the Research API for lower perceived latency.
Default tier rate limits raised from 60 to 120 RPM. No action required.
Filter results by recency: hour, day, week, month, year.
First stable release of the official Python SDK with async support and full type hints.
Cloudflare Workers, Vercel Edge, and Deno are now first-class targets.