Parsera uses large language models to understand page structure. You define attributes as natural language descriptions, like 'product price' or 'article title,' and the LLM identifies and extracts those fields from the HTML. This approach adapts to page layout changes automatically, unlike brittle selector-based scrapers that break when sites update.

Standard mode performs efficient extraction suitable for most pages. Precision mode minimizes page reduction to detect data hidden in nested HTML tags, but consumes more credits. Your agent can select the mode based on extraction complexity, using standard for straightforward pages and precision for heavily structured content.

Yes. Parsera supports proxy country routing on extraction requests. The agent can specify a proxy country like UnitedStates or Germany to access geo-restricted content. Use the Get Proxy Countries endpoint to see all available regions before configuring extractions.

You need your Parsera API key, available from the API tab on your Parsera account page. Enter it once in the Tars Tools section. The key is stored securely and enables access to all extraction, parsing, and agent management endpoints.

Absolutely. Parsera supports scraper agents and templates that define extraction attributes once and run them against any URL. The agent can execute saved templates via the Run Scraper Template endpoint, or invoke pre-built agents through the Scrape With Agent endpoint, making recurring extractions consistent and fast.

No. Parsera extraction results are fetched live during conversations and used to generate responses. Tars does not maintain a database of scraped content. Conversation logs may contain the extracted data as part of the response, but raw extraction payloads are not persisted separately.

Parsera's Python library requires coding to set up extraction scripts. Through Tars, your non-technical team members can request web data extraction conversationally. The AI agent translates natural questions into properly formatted Parsera API calls, handling attributes, modes, and proxy settings automatically.

Yes. Both the Extract Data and Scrape With Agent endpoints accept session cookies as parameters. The agent can pass authentication cookies to access login-protected pages. You configure the cookie details, and Parsera includes them in the extraction request to reach content behind authentication walls.