Claygent

What is Claygent?

Claygent is an AI research agent within Clay for gathering company, contact, and other custom go-to-market data. It searches the web for information that standard providers may not supply and can use Navigator to interact with pages, apply filters, complete search forms, click controls, and retrieve structured results.

Claygent Builder centralizes agent creation, testing, versioning, and deployment. Users describe a task in natural language, and Sculptor, Clay's AI copilot, converts the request into a production-ready prompt. Teams can add documents and business context, compare models or prompt versions on real or generated test data, roll back changes, and reuse one Claygent across multiple workflows.

How to Use Claygent

1. Describe what the agent should do in plain language and refine the generated prompt conversationally with Sculptor.
2. Upload documents, connect business context, and enable contact or job data when the task requires it.
3. Import sample rows or generate test data, compare outputs across models and prompt versions, and iterate until the results are reliable.
4. Deploy the Claygent to any Clay table and manage future prompt updates centrally across every deployment.

Core Features

• Natural-language agent building: Sculptor turns a plain-language task into a prompt that can be refined conversationally.
• Connected business context: Gives agents access to documents, company context, people, and jobs data.
• Web research and interaction: Searches for custom data and interacts with websites through Navigator.
• Free testing before scale: Tests agents on real or AI-generated data without consuming credits during testing.
• Version comparison and rollback: Compares model and prompt changes, tracks versions, and restores earlier versions.
• Central deployment management: Reuses one agent across workflows and propagates prompt updates from one place.

Use Cases

• Account research: Finds company attributes and signals that are not available from standard data providers.
• Contact research: Collects role, background, and other custom details for target people.
• Lead qualification: Researches and structures information used to score or route prospects.
• Personalized outreach preparation: Produces research inputs for relevant sales messaging.
• Structured web extraction: Applies filters or submits search forms on sites that require interaction.

Pricing

The page invites users to try Claygent for free and says Clay can be started without a credit card. It does not list Claygent-specific paid prices or production credit rates.

Frequently Asked Questions

Can Claygent interact with websites?

Yes. Navigator can apply filters, fill search forms, click buttons, and retrieve structured data.

Which AI models can Claygent use?

The page states that Claygent can run on frontier models or Claygent-tuned models and supports either user-provided or Clay-managed API keys.

Can one Claygent be reused?

Yes. A Claygent built once can be deployed across multiple workflows, with central updates propagated to every deployment.