AI search API

Real-time web search, distilled for LLMs.

ODEN turns any query into clean, citation-backed JSON your model can read. One call — no HTML noise, no scraped clutter, opt-out honored at the source.

POST /search
$ curl https://oden.samiestenbergg-6f8.workers.dev/search \
    -H "Authorization: Bearer $ODEN_KEY" \
    -d '{ "query": "what are the moons of saturn?" }'
{
  "query": "what are the moons of saturn?",
  "results": {
    "answer": "Saturn has 274 confirmed moons — more than any
              other planet. The largest, Titan, is bigger
              than Mercury and the only moon with a dense
              atmosphere.",
    "citations": [
      { "title": "Saturn's natural satellites — Wikipedia",
        "url":   "https://en.wikipedia.org/...",
        "score": 0.82 }
    ]
  }
}

Independent

Our own retrieval stack, end to end. No reseller, no middleman between your agent and the open web.

Compliant

robots.txt and EU TDM opt-out honored before a page is ever read. Citations as metadata — never reproduction.

Distilled

A synthesized answer plus ranked citation metadata. Token-dense JSON, built for context windows — not browsers.

Odin kept two ravens — Huginn and Muninn, thought and memory — who flew across the world each day and returned with everything they had seen. ODEN does the same for your model.