Learn and master Clay with Clay university

By the end of this lesson, you'll have a company search and a people search built with an AI agent, checked by hand in Clay, and saved to Audiences. The example we’ll use is a partnerships play: finding US private equity firms and the operators who work with their portfolio companies.

🔍 What's New in Search

Head to Find Leads. It now runs on a whole new companies, people, and jobs dataset, and the filters look different: everything is grouped by criteria. You still have unique company identifiers and excluded companies, plus three new things worth knowing:

1. Saved searches. You can save a search and come back to its logic later.
2. Query mode. An upgraded search that takes custom queries, part of the search beta.
3. Lookalikes and local businesses. Both live inside search now, including Google Maps as a source. A bonus lesson covers them.

You can still click filters together the old way, but we'll write the search with AI instead.

🤖 Writing the Search with an AI Agent

Query mode works with an agent skill: download the Clay search query skill, then add it to the AI tool you use. Here we use Claude.

Here's the prompt from the video:

I want your help creating a company and people search using the Clay search query skill.

For my company search, I want to look specifically for PE firms, ideally larger PE firms. We don't care necessarily about the headcount. We care more about the portfolio of companies that they manage: ideally 30+ or more, ideally larger-size companies as well, ideally in the tech sector, but not required.

On the people side, I really care about whether they have anyone on staff that deals with ops optimization, enhancement or augmentation for their portfolio companies. Even better, do they have people who specialize in revenue platform, sales platform, marketing platform, or any other type of GTM activity for their portfolio companies.

I want the PE firms and the people that we find at those PE firms to be in the United States. Go ahead and ask me any other questions that'll help you help me before getting started.


Remember to ask the AI to ask you questions before it starts. In our example, Claude came back with four which dramatically changed the query:

Why run both a company search and a people search? It's still the best way to resolve entities across a one-to-many relationship: one firm, many operators.

This is also a partnerships play, not a direct sales one. GTM engineering isn't just cold outbound. The same tools run inbound, CRM enrichment, direct mail, and plays like this one.

📋 The Two Queries

Claude returned two queries that read a lot like SQL.

Company search: US PE and growth-equity firms with a portfolio-ops or GTM bench.

select from companies
where
 industry in ("Venture Capital and Private Equity Principals", "Investment Management")
 and (
   description contains ("private equity", "growth equity", "buyout")
   or products_and_services is_similar_to ("private equity", "growth equity")
 )
 and locations.any(country_name = "United States")
 and people.exists(
   is_current = true
   and person.location_country = "United States"
   and seniority in ("Partner", "C-suite", "VP", "Director", "Head")
   and (
     job_title is_similar_to ("Operating Partner", "Portfolio Operations", "Value Creation", "Operating Executive", "Operational Excellence")
     or (
       job_title contains ("go-to-market", "GTM", "revenue", "sales", "marketing", "commercial")
       and (
         job_title contains ("portfolio", "value creation", "operating")
         or description contains ("portfolio companies", "portfolio company", "value creation", "revenue platform", "sales platform", "marketing platform")
       )
     )
   )
 )


People search: those operators at those firms.

select from people
where
 location_country = "United States"
 and experiences.any(
   is_current = true
   and company.industry in ("Venture Capital and Private Equity Principals", "Investment Management")
   and (
     company.description contains ("private equity", "growth equity", "buyout")
     or company.products_and_services is_similar_to ("private equity", "growth equity")
   )
   and company.locations.any(country_name = "United States")
   and seniority in ("Partner", "C-suite", "VP", "Director", "Head")
   and (
     job_title is_similar_to ("Operating Partner", "Portfolio Operations", "Value Creation", "Operating Executive", "Operational Excellence")
     or (
       job_title contains ("go-to-market", "GTM", "revenue", "sales", "marketing", "commercial")
       and (
         job_title contains ("portfolio", "value creation", "operating")
         or description contains ("portfolio companies", "portfolio company", "value creation", "revenue platform", "sales platform", "marketing platform")
       )
     )
   )
 )


Claude also flagged what the queries can't capture: Clay's search data has no portfolio size and no portfolio sector, so it suggested a Claygent enrichment afterward. That's exactly what we build in the workflows lesson.

✅ Checking the Search in Clay

Pasting the query is not the end of the job. Code is hard to read, and you need to understand what the search is actually doing, the same way you'd read a pull request before merging AI-written code.

1. Paste the company query into query mode.
2. Open the filter screen and read it back: descriptions containing private equity, the right industries, more than one senior person in the US.
3. Look for surprises. Some venture firms came through. That's fine for this play, since they also have portfolio companies and platform teams.
4. Fix the nuance. The job-title filter was set to match all titles. Switching it to any broadened the list from about 1,000 companies to about 3,000.

These small calls are where list quality lives, whether you build in code, in Clay, or anywhere else. You don't need to memorize where every button is, but you do need two things:

1. Taste
2. The ability to troubleshoot

To narrow further, you could drop the VC firms and run a separate play for them.

💾 Saving to Audiences

1. Click Continue.
2. Choose Max available.
3. Save to Audiences. You could enrich the companies here too, but we'll do all of that in a workflow from the CLI.

From your prompt to the saved list: one prompt, four answers, two queries, one list of US PE firms living in Audiences.

🚀 What's Next

You can now write a company and people search with an AI agent, read it back in Clay, and save the results to Audiences.

Next, we'll look at Audiences itself and turn those results into a segment you can act on.