Skip to main content
Mythos

Claygent is a feature of 📝Clay that turns natural-language prompts into AI research agents — small programs that crawl the web, navigate sites, and enrich records across a Clay workspace.

It collapses the friction of building outbound enrichment: what previously required hand-coded scrapers, brittle selectors, or twenty minutes of formula construction now reduces to a one- or two-sentence prompt, with free testing on real data before any credits are consumed.

Built for 📝Go-To-Market (GTM), sales-ops, and revenue-ops teams running outbound at scale — anyone using Clay to enrich, qualify, or research accounts and contacts.

Claygent was a step-change unlock in our usage of Clay at 📝One Inc — template building that used to take twenty or thirty minutes of formula and selector work now lands in a one- or two-sentence prompt.

Key Capabilities

  • Natural-language agent building — convert plain-language task descriptions into production-ready research agents that hit live data sources without writing prompts or selectors by hand.
  • Web research and navigation — agents scour the open web and use 📝Navigator to apply filters, fill forms, paginate, and extract structured data from sites that resist scraping.
  • Free testing — run agents against real workspace data and refine prompts iteratively without burning a single credit until you're ready to scale to a full production deployment.
  • Centralized deployment — build an agent once and deploy it across every workflow in the workspace, with edits to the source propagating globally to every usage in real time.
  • Multi-model support — run agents on frontier models or Clay-tuned variants, with bring-your-own-key for cost control or Clay-managed billing for instant plug-and-play setup.

Getting Started

  1. Sign in to Clay or sign up at app.clay.com/signup.
  2. Open any Clay table and add a Claygent column.
  3. Describe the task in plain language — 📝Sculptor turns it into an agent prompt.
  4. Test on a few rows for free before committing credits to a full run.
  5. Deploy the agent across your workflows; updates propagate to every usage.

Related

Contexts

Created with 💜 by One Inc | Copyright 2026