Structured Extract

01 / Contract

One request. A usable object.

The schema describes the output you need. Purify obtains and cleans the page, then applies that contract with your LLM provider.

Request / cURL

curl -X POST https://purify.verifly.pro/api/v1/extract \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "url": "https://example.com/article",
    "schema": {
      "title": "string",
      "author": "string",
      "topics": ["string"]
    },
    "llm_api_key": "YOUR_LLM_KEY"
  }'

Response / JSON

{
  "data": {
    "title": "A practical data systems guide",
    "author": "A. Example",
    "topics": [
      "web data",
      "search",
      "agents"
    ]
  }
}

Example content is illustrative; target-page results vary.

02 / Process

Structure follows purification.

01 — INPUT

URL + schema

Describe the fields and types the downstream system expects.

02 — REFINE

Clean content

Purify reduces page furniture before interpretation begins.

03 — OUTPUT

Structured JSON

A predictable object can move directly into search, storage, or agent tools.

03 / Uses

For systems that need more than prose.

01 / Retrieval

RAG pipelines

Turn documentation, articles, and knowledge bases into predictable objects before chunking and indexing.

02 / Action

AI agents

Give an agent a typed result instead of asking it to repeatedly reason over navigation, scripts, and page furniture.

03 / Operations

Data enrichment

Add selected web fields to a product catalogue, research corpus, or internal record without maintaining selectors.

04 / BYOK

Your model key. Your choice.

Structured extraction requires a model to interpret the requested schema. Supply the compatible provider key in the request, and control the model relationship yourself.

Start with Purify View pricing