Claygent: Your All-Knowing Digital Assistant

Imagine having a tireless, incredibly knowledgeable assistant at your fingertips, ready to scour the web for any information you need. That's Claygent in a nutshell. It's designed to access and retrieve virtually any publicly available data point on the internet, with the exception of information hidden behind paywalls or password-protected areas.

What makes Claygent truly unique is its ability to harness both the vast expanse of web data and the analytical power of AI. However, it's important to remember that while Claygent has access to an enormous amount of information, its effectiveness relies heavily on the context and instructions you provide. Mastering the art of prompting is key to unlocking Claygent's full potential.

In this lesson, we’ll show you how to get started with Claygent.

How To Use Claygent in Clay

Watch the guided demo to learn how to use Claygent, or skip ahead for a step-by-step breakdown.

Step 1: Getting Started With Claygent

To begin your Claygent journey, click on “Add enrichment” in the top right corner of your table, select “AI” from the sidebar navigation, then click “Use AI.” That’s it! The Claygent modal will pop out.

Step 2: Selecting A Model

When using Claygent, you have the flexibility to choose from different AI models, each with its strengths:

1. Claygent Helium: The best Claygent model at its price-performance, outperforming Haiku and GPT 4o-Mini.
2. Claygent Neon: While it may not have the same level of reasoning capabilities as some other models, it excels in answer formatting and data extraction. Performs well on more complex tasks at a reasonable price point, outperforming Sonnet 3.5
3. Claygent Argon: Strongest overall model for deep research and complex analysis, significantly outperforming Opus and GPT 4o.
4. GPT-4 and Claude Opus: These models offer enhanced reasoning capabilities. While they don't provide the same level of answer formatting as Claygent models, they can be useful for testing data accuracy or handling more complex analytical tasks.

Selecting the right model is more of an art than a science, and we encourage you to experiment to find the best fit for your specific needs.

Step 3: Templates & Prompts

We've made it easy to hit the ground running by providing a library of preset templates for common tasks. These templates are an excellent starting point, especially if you're new to working with AI agents.

These templates provide pre-built prompts for common web research tasks, such as:

• Finding a company's 10K report
• Extracting recent fundraising information
• Pulling contact details from LinkedIn
• Identifying a company’s ICP
• Finding number of locations
• And much more

If no template fits your needs, create a custom Claygent prompt using the S.P.I.C.E. framework and meta-prompter principles covered in the previous lesson.

Step 4: Configuring Outputs

When scraping multiple data points using a single Claygent prompt, define column outputs to ensure structured and well-organized data. This step is crucial for breaking down complex results into manageable, actionable insights.

Here’s how it works:

1. Identify the Data Points Needed: Determine the specific information you want Claygent to return, such as funding stage, amount raised, or lead investors.
2. Define Column Outputs: In the Claygent pop-out modal, define column outputs for each respective piece of information. For example, you can create a column output for funding stage, amount raised, or lead investors. 
3. Instruct Claygent: In the prompt, tell Claygent which information you want to send to which column output you just defined. For example, tell Claygent “Output the funding stage to the ‘Stage’ field; output the amount raised to the ‘Amount’ field.”

‍💡Tip: Before applying Claygent to a large dataset, run a test on a few rows, review the structured output, and make adjustments if necessary.

Practical Applications: From LinkedIn Profiles to Thought Leadership

Let's explore some real-world applications of Claygent to illustrate its versatility:

1. LinkedIn Profile Enrichment: With just a few input variables, Claygent can locate an individual's LinkedIn profile and extract valuable information such as work experience, awards, and summary.
2. Social Profile Analysis: Once you have the LinkedIn data, you can use Claygent  to summarize the information and highlight the three most unique or interesting insights about the individual.
3. Thought Leadership Research: Claygent truly shines when tasked with finding thought leadership content. You can instruct it to search for blogs, interviews, podcasts, and articles related to a specific person. Using the Neon model, you can extract multiple pieces of research into separate outputs, making it easy to organize and analyze the findings.
4. Content Analysis: For each piece of research found, you can run another Claygent call to extract metadata and key excerpts. This might include classifying the content type (blog, podcast, interview, etc.), identifying the author, summarizing main points, and pulling relevant quotes or snippets.
5. Personalized Outreach: All this rich, personalized data can then be used to craft highly targeted and engaging outreach messages, significantly improving your communication strategies.

Congrats! 

That’s a wrap on Claygent, Clay’s infamous web research agent. 

Next up, we’ll explore A.I. formulas and how you can use them to structure and normalize data automatically.

‍

‍