Skip to content

Scrape Public LLM Shared Chats Through One API Call

Send a public shared-chat URL and receive rendered HTML or structured JSON.

OUTPUT FORMATSHTMLJSONMarkdown

The service runs JavaScript, rotates proxies and handles anti-bot checks, so you do not need to operate a browser fleet or source proxy providers yourself. A typical result includes prompt text, response text, citation URLs, model variant and the turn-by-turn structure of the conversation.

Three practical uses for shared-chat data

Most teams collecting this data want one of three outcomes. First, conversation snippets for competitor analysis and positioning. Second, brand-mention monitoring across public chats to see where AI users discuss a product or category. Third, citation tracking: the source URLs and references a model cites, which helps map what a model relies on and check factual grounding. Decide which outcome you need before choosing extraction rules, because the fields you request should match the decision you will make with them.

How a request flows, step by step

Use this sequence as a pre-flight checklist:

  • Confirm the URL is a public shared-chat page, not post-login content, and check the target site's terms.
  • Create an account and copy your API key; starter credits let you test without a card.
  • Pick the SDK for your stack, or use cURL, so each page is a single function call.
  • Send the URL by GET or POST and choose rendered HTML for inspection or structured JSON for pipelines.
  • Add parameters only when needed: render_js for JavaScript-heavy pages, stealth_proxy for CAPTCHA challenges, country_code for geo-targeting, and ai_extract_rules for natural-language extraction.
  • Validate prompt text, response text, citations and turn order in the first responses before scaling volume.

A useful worked example: for citation tracking, request structured JSON with fields for each cited URL and its surrounding response text, then group URLs by frequency. For brand monitoring, extract response text and search it for your brand, product and competitor names. For snippet collection, keep the full turn structure so prompts and answers stay paired.

HTML, JSON and extraction rules: which to choose

Two extraction paths cover most needs. extract_rules maps CSS selectors to named fields when you know the page structure. ai_extract_rules takes a natural-language specification when the layout is less predictable. Screenshots, JavaScript scenarios, Markdown output and no-code automation are available as companion capabilities when a plain HTML or JSON response is not enough. Keep the distinction clear: rendered HTML preserves the page for your own parsing, while extraction rules return only the fields you defined.

Pricing and plan selection

Plans are separated chiefly by two variables: the monthly API credit allowance and the number of requests that may run concurrently. Credits expire after one month and do not carry over, and each request begins at a base credit cost that increases when options such as JavaScript rendering or premium and stealth proxies are enabled. Higher tiers add priority email support, a dedicated account manager and team management with per-user quotas. Listed prices exclude VAT. Because commercial terms can change, treat the figures below as a worksheet to complete from the current published pricing rather than as a quotation: monthly price = ____; included credits = ____; concurrency = ____; credit cost per request with JavaScript rendering = ____; with premium proxy = ____; with stealth proxy = ____. Compare those inputs against your expected monthly page volume before choosing a tier.

Compliance, blocking and performance notes

Compliance and performance are best evaluated as questions rather than assumed guarantees. On data handling, ask whether the provider collects or stores personal data from scraped pages unless the user explicitly requests it, and whether GDPR and CCPA alignment is documented. On blocking, ask how proxy rotation and headless-browser rendering are used to reduce IP-based blocking by mimicking ordinary browsing behaviour. On scale, ask what browser and proxy capacity the platform is designed to handle and what response times are typical, including for JavaScript-heavy pages. A reasonable evaluation is to run a small sample of public shared-chat URLs, record success rate and latency, and confirm the answers in the provider's current documentation before relying on them in production.

Frequently asked questions

Do I need an account to scrape LLM-generated content?

No separate account for the target chat service is needed. You create an account for the API itself, receive an API key and starter credits, then send the public shared-chat URL.

Is scraping LLM-generated content legal?

Public-facing pages are usually scrapeable, but check the target site's terms before running a job and stay away from content behind a login.

What's the rate limit for scraping LLM-generated content?

Limits are set by your plan's concurrency rather than a per-target rate. If a target throttles aggressive patterns, space requests out or spread them across sessions.

How do I get structured JSON from LLM-generated content?

Supply extract_rules mapping CSS selectors to field names, or ai_extract_rules with a natural-language specification. The API returns the extracted fields as parsed JSON.

Which programming languages can I use to scrape LLM-generated content?

Any common stack works. SDKs are published for Python, Node.js, Go, PHP, Ruby and Java, and plain cURL is also supported.

How much does it cost to scrape LLM-generated content?

Cost has two parts: the monthly plan and the credit charge per request. Plans bundle a monthly credit allowance with a concurrency limit, and the per-request charge rises when options such as JavaScript rendering or premium and stealth proxies are enabled. To budget, fill in the current published figures: monthly price = ____; included credits = ____; concurrency = ____; credit cost per request with JavaScript rendering = ____; with premium proxy = ____; with stealth proxy = ____. Multiply your expected monthly request count by the applicable per-request charge, compare the total with the included credits, and confirm any starter-credit or no-card terms directly with the provider before signing up.