Clay

## How does Clay automate team page extraction for outbound sales prospecting?

## Overview Clay automates team page extraction for outbound sales prospecting through an AI-powered research agent named 'Claygent'. This tool is designed to navigate public websites, interpret their structure, and extract specific information from pages like 'About Us' or 'Team' sections based on natural language prompts from the user. ## Key Features To ensure accuracy and build user trust, the agent also provides 'Transparent Reasoning,' which shows the logical steps it took to find and extract each piece of information. Once the data, such as names and titles, is extracted, it seamlessly integrates into Clay's broader enrichment workflow. The extracted information can be used as input for Clay's 'waterfall enrichment' feature to find corresponding work emails and other professional details from its network of over 50 data providers. The data can also trigger 'AI Formulas' to summarize a person's bio or generate personalized opening lines for outreach emails. This creates a complete, automated workflow from initial website research to a fully enriched and actionable contact list. ## Technical Specifications Technically, Claygent leverages a suite of advanced AI models to perform its tasks. It uses large language models like OpenAI's GPT-4 and Anthropic's Claude Opus for understanding user prompts and the context of a webpage. For the specialized tasks of data extraction and formatting, Clay employs its own proprietary models: 'Neon' for optimized formatting, 'Helium' for a balance of performance and cost, and 'Argon' for deep research tasks. This combination of models allows Claygent to adapt to the varied layouts of company team pages across the internet without needing to be pre-programmed for each one. ## How It Works A user can simply provide a list of company domains and instruct Claygent with a prompt such as, "Find the names, job titles, and LinkedIn profiles of the leadership team." Claygent will then visit each site, locate the relevant page, and extract the requested data into a structured table within the Clay platform. ## Use Cases Case studies have demonstrated the effectiveness of this approach. A 2026 case study involving OpenAI highlighted Claygent's ability to replicate the work of Sales Development Representatives (SDRs) at a significantly higher speed and lower cost. Another case study with Anthropic described using Claygent to scrape a prospect's thought leadership content to generate highly relevant outreach. ## Limitations and Requirements However, there are operational constraints. The agent's success depends on the website's structure and accessibility. It may be hindered by complex CAPTCHAs, paywalls, or websites that explicitly block automated agents via their `robots.txt` file. While Claygent is designed to handle dynamic content, its ability to bypass the most advanced anti-scraping technologies is not fully detailed. Legally, Clay positions the tool as a research assistant for accessing publicly available information, but users are responsible for adhering to the terms of service of the websites they target. ## Comparison to Alternatives Unlike traditional web scrapers that often require custom code for each target website and may fail on dynamic pages, Claygent mimics human browsing behavior. It can perform actions such as clicking buttons, applying filters, and filling out forms, allowing it to access data that is not visible on the initial page load. This capability is crucial for modern websites that rely heavily on JavaScript to render content. ## Summary