Skip to content

Collect Public Microsoft Copilot Chats Through an API

Use one API call to retrieve publicly shared Microsoft Copilot conversations for citation tracking and brand-mention monitoring, without operating browsers, proxy rotation, or anti-bot handling yourself.

OUTPUT FORMATSHTMLJSONMarkdown

Each call returns fresh rendered data for the URL you supply.

What to collect from public shared chats

Publicly shared chats are useful for three distinct jobs. First, query research: read how people phrase requests, which capabilities they try, and which conversations are shared most. Second, brand monitoring: find where your brand, product, or a competitor is named in shared chats. Third, citation tracking: collect the sources, documents, and websites Copilot recommends inside those answers. Keep these jobs separate in your pipeline — a spike in mentions is not the same signal as a spike in citations, and neither proves how private or unshared conversations behave.

Field map for a shared chat

A public shared chat can expose prompt text, response text, citation URLs, suggested follow-ups, and the turn-by-turn structure of the conversation. Treat that list as your field map when designing extraction: capture each turn in order, attach citations to the answer they support, and preserve follow-up suggestions as separate fields rather than flattening them into the response text. That structure lets you answer practical questions later — which prompts lead to which sources, and where a brand appears in the flow of a conversation.

Four-step workflow

  1. Create an account and copy your API key. 2. Install the SDK for your stack — Python, Node, Go, PHP, or Ruby — so each scrape is a single function call. 3. Send the Copilot URL with your API key as a parameter; the renderer loads the page in a real browser, rotates the IP, and returns the response. 4. Add advanced parameters only when needed: render_js for JavaScript execution, stealth_proxy for CAPTCHA-heavy targets, country_code for geo-targeting, and ai_extract_rules for natural-language extraction. Test first in the dashboard request builder, then move the generated snippet into your codebase.

Choosing an extraction mode

Choose extract_rules when the page structure is stable and you want precise CSS or XPath targeting. Choose ai_extract_rules when the fields are easier to describe than to select, accepting that description-led extraction needs validation against the rendered page. Add screenshots when you need visual evidence alongside data, Markdown output when you need clean LLM-ready text, and JavaScript scenarios when content only appears after clicks, scrolling, or typing. The decision rule is simple: deterministic selectors first, AI extraction for flexible fields, and browser scenarios only when interaction is required to reveal the content.

Blocking, scale, and privacy checks

Proxy rotation and headless-browser rendering reduce IP-based blocking by mimicking ordinary browsing. For scale, the platform is described as using thousands of headless browsers with rotating proxies, with typical results in 1–5 seconds even on JavaScript-heavy pages. On privacy, the service states it does not collect or store personal data from scraped sites unless a user explicitly requests it. Regardless of tooling, review the target site's terms and avoid post-login content.

Plan comparison (owner-confirmed)
Plan tierMonthly priceAPI creditsConcurrent requestsSupport and team features
Hobby{{OWNER_PRICE_HOBBY}}{{OWNER_CREDITS_HOBBY}}{{OWNER_CONCURRENCY_HOBBY}}{{OWNER_SUPPORT_HOBBY}}
Freelance{{OWNER_PRICE_FREELANCE}}{{OWNER_CREDITS_FREELANCE}}{{OWNER_CONCURRENCY_FREELANCE}}{{OWNER_SUPPORT_FREELANCE}}
Startup{{OWNER_PRICE_STARTUP}}{{OWNER_CREDITS_STARTUP}}{{OWNER_CONCURRENCY_STARTUP}}{{OWNER_SUPPORT_STARTUP}}
Business{{OWNER_PRICE_BUSINESS}}{{OWNER_CREDITS_BUSINESS}}{{OWNER_CONCURRENCY_BUSINESS}}{{OWNER_SUPPORT_BUSINESS}}
Business +{{OWNER_PRICE_BUSINESS_PLUS}}{{OWNER_CREDITS_BUSINESS_PLUS}}{{OWNER_CONCURRENCY_BUSINESS_PLUS}}{{OWNER_SUPPORT_BUSINESS_PLUS}}
Enterprise 14{{OWNER_PRICE_ENTERPRISE_14}}{{OWNER_CREDITS_ENTERPRISE_14}}{{OWNER_CONCURRENCY_ENTERPRISE_14}}{{OWNER_SUPPORT_ENTERPRISE_14}}
Enterprise 24{{OWNER_PRICE_ENTERPRISE_24}}{{OWNER_CREDITS_ENTERPRISE_24}}{{OWNER_CONCURRENCY_ENTERPRISE_24}}{{OWNER_SUPPORT_ENTERPRISE_24}}
Enterprise 41{{OWNER_PRICE_ENTERPRISE_41}}{{OWNER_CREDITS_ENTERPRISE_41}}{{OWNER_CONCURRENCY_ENTERPRISE_41}}{{OWNER_SUPPORT_ENTERPRISE_41}}
Enterprise 69{{OWNER_PRICE_ENTERPRISE_69}}{{OWNER_CREDITS_ENTERPRISE_69}}{{OWNER_CONCURRENCY_ENTERPRISE_69}}{{OWNER_SUPPORT_ENTERPRISE_69}}
Enterprise 120{{OWNER_PRICE_ENTERPRISE_120}}{{OWNER_CREDITS_ENTERPRISE_120}}{{OWNER_CONCURRENCY_ENTERPRISE_120}}{{OWNER_SUPPORT_ENTERPRISE_120}}

Owner-confirmed template only: {{OWNER_PLAN_PRICES_AND_CREDIT_ALLOWANCES}}. Do not publish until the owner confirms current prices, credit allowances, concurrency, support inclusions, VAT treatment and cancellation terms. Credits are described as valid for one month without rollover; JavaScript rendering and premium proxy requests carry additional per-request credit costs under {{OWNER_CREDIT_COST_RULES}}.

Frequently asked questions

Is scraping Copilot legal?

Pages that are publicly reachable are generally the ones that can be collected, but you should still check the target site's terms before running a job and stay away from content behind a login.

What data can I extract from Copilot?

Anything visible on the public page can be returned. For repeatable field-level output, use CSS-based extract_rules; for looser, description-led extraction, use ai_extract_rules and request structured JSON.

How do I start scraping Copilot?

Create an account, copy your API key, then send a GET request containing the target URL. The service renders the page and rotates proxies before returning data. A trial credit allowance is described in the owner-confirmed offer below.

Will I get blocked when scraping Copilot?

Requests are routed through rotating residential proxies and a real headless browser so they resemble ordinary browsing. When a target is more aggressive, turn on stealth_proxy for that request.

Can I scrape Copilot without writing code?

Yes. The dashboard request builder accepts a URL and parameters, then produces a ready-to-copy snippet for Python, Node, Go, PHP, Ruby, or cURL.

How do I handle JavaScript when scraping Copilot?

Set render_js=true. The page runs in a real headless browser on the server, JavaScript executes there, and you receive the HTML after rendering.