Search · Extract · Crawl · Research — one API

The search & research API for AI agents

Search, extract, crawl & research the live web in one API — with best-in-class reranking so your AI reads the right sources first. Drop-in compatible, ready in minutes.

Start free Read the docs

Live demo · rate-limited · no key required.

0+
Search sources
aggregated into one API
4-in-1
One API
search · extract · crawl · research
0.0s
Warm response
smart caching
1 line
To migrate
swap one base URL

Works with the tools you already use

LangChain LlamaIndex OpenAI Anthropic Vercel AI SDK CrewAI LangGraph Haystack Python TypeScript cURL n8n LangChain LlamaIndex OpenAI Anthropic Vercel AI SDK CrewAI LangGraph Haystack Python TypeScript cURL n8n
Why it's better

Best results first,
not a wall of links.

Most search APIs hand your AI a raw, noisy list. Crisps reads and ranks every result for how well it truly answers you — so the sources that matter rise to the top.

raw web resultsunsorted
  • 1Reddit: 'is fusion even real?'—
  • 2ITER — commissioning milestone—
  • 3SEO listicle: 10 energy startups—
  • 4Nature: net-energy, peer-reviewed—
  • 5Commonwealth Fusion — SPARC—
How it works

A two-stage pipeline for signal, not noise

Snippet-rerank the candidate pool, extract only the winners, then rerank each page's chunks — so you fetch less and rank better.

  1. 01

    Query

    Your agent asks in natural language.

  2. 02

    Aggregate

    70+ engines across the web, deduped.

  3. 03

    Extract

    Clean markdown; PDF + headless-browser fallback.

  4. 04

    Rerank

    Cross-encoder → true 0–1 relevance + top chunks.

  5. 05

    Respond

    Drop-in JSON, cached for ~0.0s warm.

Migrate in minutes

Drop-in for the search API
you already use.

Crisps speaks the same contract as the leading cloud search API. Change one line of config and your agents keep working — now with sharper, reranked results.

  • Same request & response shape as the popular search API
  • Swap one base URL — your agent code stays identical
  • Sharper, reranked results out of the box — no tuning
.env 
# before — another search API
SEARCH_API_BASE=https://api.<vendor>.com

# after — Crisps (same contract)SEARCH_API_BASE=https://api.criiiiissss.shop
python 
client = Crisps(base_url=SEARCH_API_BASE)
res = client.search("deepseek v3 moe", search_depth="advanced")
# → results[].{title, url, content, score, raw_content}
Benchmarks

How Crisps compares

Head-to-head with the leading cloud search API on the same real queries — better ranking, more tools in one call, faster repeats.

Metric
Crisps
Leading cloud API
Relevance ranking
cross-encoder, true 0–1
opaque black box
Tools in one API
search · extract · crawl · research
search + paid add-ons
Warm response
0.0s (cached)
~0.8s
Extraction coverage
~100% (PDF + JS)
varies
Migration effort
one base URL
rewrite integration
Trained on your data
never
varies by vendor

Warm responses are served from cache; ranking uses a cross-encoder for true query relevance.

Give your agents the web.
Answers, not noise.

Start free in minutes — no credit card. Drop-in compatible, ranked for quality.

Start free Read the docs