Why a normal search only finds the obvious names
When you need every agency in a niche, every tool in a software category, or every expert on a topic, a search engine gets you started and then stalls. Search results are ranked by popularity, so the first page is full of roundup articles that repeat the same ten well-known names. The small agency that does exactly what you described, the tool that launched last spring, and the practitioner who writes the best posts on the subject rarely appear there.
Building a complete list by hand is slow for three reasons:
- You have to search the same idea in several different ways, because a company that calls itself a "growth studio" won't show up when you search for "marketing agency."
- You have to open each candidate's website or profile to see whether it really fits, because a directory listing or a passing mention isn't proof.
- You have to merge the results from every search, remove the duplicates, and keep track of who you already checked.
This is the kind of work an AI agent handles well. It can run many searches in parallel, read every candidate's own pages, apply the same test to each one, and keep a record of why each name made the list.
How Kite finds every company or person that fits a description
Kite starts by turning your description into a test that a company or person either passes or fails. It then looks at how that test is built. When the conditions are facts that business databases store, Kite filters those databases and gets the whole matching group in one pass. When the conditions are things no database records, Kite searches the public web from several angles to gather candidates.
Every candidate from either route gets checked against every condition, and only the ones that pass go on the list. Kite then tells you how complete the list is and what it would take to widen it. Once you're happy with the list, Kite can put it to work by saving it to your CRM, finding contact details, or drafting outreach.
- Turn the description into a test every candidate must pass
- Search business databases when the description is made of filters
- Search the public web when no database stores the description
- Check every candidate before it goes on the list
- Know how complete the list is
- Put the list to work
Turn the description into a test every candidate must pass
A complete list is only possible when "fits the description" has a clear answer, so Kite first breaks your description into separate conditions. Suppose you ask for "every agency that runs paid social for dental practices in the US." Kite reads that as four conditions: the company is an agency, it runs paid social campaigns, dental practices are among its clients, and it operates in the US.
Kite settles the details that your words leave open. It decides whether "dental practices" includes orthodontists, whether a freelancer counts as an agency, and whether "runs paid social" means the agency lists it as a service or shows client work to prove it. When a choice would change the list a lot, Kite asks you before it starts. It also uses what it already knows about your business to set exclusions, such as leaving out your current customers and competitors when the list is meant for prospecting.
Search business databases when the description is made of filters
When your conditions are facts that business databases store, Kite filters those databases directly, because one filtered search returns the whole matching group with the facts already attached. Kite's built-in sources include Apollo for companies, people, and work emails, and Crustdata for funding, hiring, leadership, and technology use. You don't need an account with either one.
Company searches can filter by industry, size, and location, with further filters for headcount growth, funding, and technology use. People searches can filter by title, function, seniority, location, and current employer. If you already have a few companies that fit, Kite can also find others similar to them.
If you have a DataForB2B account, connect it and Kite can search there too. DataForB2B adds filters for things such as a person's skills, past employers, and investors, and it can find people similar to a profile you give it.
Playbooks for this step
- How to build a qualified prospect list with ApolloTurn your ideal customer profile into Apollo's company and people filters, then check each match for current role, fit, and a reason to reach out before it goes on the list.
- How to find LinkedIn prospects with CrustdataTurn the audience you have in mind into filters on Crustdata's person and company databases, then confirm each match still fits and look up a work email.
Show 2 more playbooks for this step
- How to find people and companies with DataForB2BTurn a description of your audience into company and people searches in DataForB2B, then get a checked and scored list with verified work emails.
- How to find companies similar to your best customers with CompanyEnrichStart from the companies you already sell to and get a list of similar companies, with the accounts you already know removed and every match scored against your ideal customer profile.
Search the public web when no database stores the description
Many useful descriptions can't be expressed as database filters, so Kite searches the public web to find who fits them. "Tools that turn podcast episodes into blog posts," "agencies that specialize in Shopify migrations," and "people who have written in depth about pricing for developer tools" are all examples. No company database has a field for any of them.
For these, Kite uses Exa, a built-in source for searching the public web, to discover companies, people, articles, and other entities that match your description. Kite also runs its own searches using different vocabulary, because the same kind of company describes itself in different ways. A search for "AI meeting notes apps" and a search for "call transcription software" will surface different products that belong on the same list.
For lists of people, the web route finds experts by what they have done in public. Kite looks for people who have written about the topic, spoken at conferences, appeared on podcasts, or been quoted in the press. It can also search public conversations on X and Hacker News to find the practitioners other people turn to.
Playbooks for this step
- How to research target accounts with ExaUse Exa to find and research companies that match a description, with a source for each finding.
- How to build a media list of journalists who cover your industryFind the writers who cover your topic from the articles published on it in recent months, check that each one still covers that beat, and keep a short list with the evidence for every name.
Check every candidate before it goes on the list
A name only goes on the list after Kite has checked it against every condition, because candidates from any search include near misses. A web search for dental marketing agencies also returns agencies that wrote one blog post about dentists, and a database search for "Head of Marketing" also returns people who left the company last quarter.
For each candidate, Kite records whether each condition was met, the reasoning behind that decision, and the page that supports it. That might be the agency's own services page, the tool's pricing page, or the article an expert wrote. Candidates that fail a condition are left off the list, but Kite keeps the reason so you can see what was excluded and why.
Kite also merges what every search found. The same company can appear in a database result and a web result under slightly different names, so Kite combines them into a single entry with everything it learned from both.
Know how complete the list is
Kite tells you how much of the market your list covers, so you know whether you have every match or only the first batch. Alongside the list, it reports how many candidates it found, how many passed the test, and which searches it ran to find them.
When Kite can tell that the list isn't complete, it says so directly. For example, it might report that the database search covered US agencies thoroughly, but that agencies without a website listing their services can only be found through referrals or directories it hasn't searched yet. Kite then asks whether you want to widen the search, narrow the description, or stop with what you have.
Markets change, so a list you build today will miss next month's new entrants. You can ask Kite to repeat the search on a schedule and point out the names that weren't on the previous list.
Put the list to work
A list is usually the start of a job, so Kite can carry it into the work that follows. With a connected CRM or spreadsheet, it writes the companies and people there. It can find professional emails and LinkedIn profiles for the people on the list, score each entry from one to five against your ideal customer profile, and draft personalized outreach that you approve before anything is sent.
The list can also feed other research. A list of every tool in your category becomes the starting point for a competitive analysis. A list of experts becomes a list of podcast hosts to pitch, writers to brief, or speakers to invite. Kite saves useful findings in your team's knowledge base so later work can build on them.
Playbooks for this step
- How to add qualified leads to HubSpotCheck each lead's fit, find the records that already exist, and write contacts and companies into HubSpot with a lifecycle stage, an owner, and the reason sales should call.
- How to save a qualified lead list to Google SheetsGive each lead one row with their role, company, LinkedIn profile, verified work email, a fit score with its reason, and a link to the source that proves it.
Show 2 more playbooks for this step
- How to score B2B leads against your ideal customer profileTurn your profile into criteria you can check, confirm each lead is still current, and give every lead a fit score with a one-line reason.
- How to write a personalized cold email from prospect researchOpen each email with a recent, verified fact about the recipient that explains why you're writing now, and approve every draft before it goes out.
What you get
A checked list
Every company or person who passed the test, with a website or profile link and the key facts that made them qualify.
A source for every match
Each entry shows, condition by condition, why it fits and the page that proves it.
The near misses
The candidates Kite left out and the condition each one failed, so you can bring any back if your definition changes.
A coverage report
How many candidates Kite found, how many qualified, which searches it ran, and which parts of the market it hasn't covered yet.
The next step, ready to go
Optional contact details, fit scores, CRM records, and outreach drafts for the people and companies on the list.
How to get started
Tell Kite what you're looking for in plain words, in the Kite web app or in Slack. You might ask, "Find every agency in the US that runs paid social for dental practices," or "Build me a list of everyone who has written seriously about usage-based pricing in the last two years."
Kite is an AI marketer. It researches your business, builds a growth strategy, and ships the work that brings you new customers. It already knows your customers, competitors, and ideal customer profile from that research, so a short description is usually enough. Kite replies with any choices it needs you to make, and once you settle them, it starts searching.
You don't need to connect anything to begin. Apollo, Crustdata, and Exa come built into Kite. Connect DataForB2B if you already use it and want its filters in the search, and connect your CRM or a spreadsheet when you want the finished list saved there.
Integrations used to find companies and people by description
Kite combines two kinds of sources for this work. Business databases answer descriptions made of filters, and web search answers descriptions that no database records.
Search business databases for people and companies

Apollo is built into Kite. Kite uses it to find the companies that match your filters and the right people inside them, then add verified work emails and phone numbers to the people who qualify.

Crustdata is built into Kite, so you can use it without your own key. Kite searches its company and person databases, filters companies by headcount growth, and pulls each contact's employment history and work email.
Connect your DataForB2B account so Kite can search its companies and professional profiles by filters, plain-English descriptions, or similarity to a profile you already have, and add verified work emails to the people who qualify.
Find companies, tools, and people on the public web
Common questions
Can ChatGPT find every company that fits a description?
ChatGPT can give you a useful starting list, but it isn't built to produce a complete or checked one. It answers from what it learned in training plus a small number of web searches, so it tends to name the same well-known companies a search engine would. Kite runs many searches across business databases and the public web, checks each candidate against your conditions with a source, and reports how much of the market it covered.
How do I build a list of experts on a topic?
Start by defining what counts as expertise for your purpose, such as having published original work on the topic, having spoken about it at an industry event, or having practiced it at a known company. Then look for public evidence of that expertise, because a job title alone rarely proves it. Kite searches articles, conference speaker lists, podcast appearances, press quotes, and discussions on X and Hacker News, then keeps each expert's strongest piece of evidence beside their name.
How long does it take to build a complete list?
A list drawn from business database filters is usually ready within minutes. A list that depends on searching the public web and checking each candidate takes longer, often tens of minutes for a broad description, because Kite reads each candidate's own pages before it decides. Kite tells you when it starts and brings the results back to the conversation when the work is done.
Do I need my own data subscriptions?
No. Apollo, Crustdata, and Exa are built into Kite, so you can build a full list without any account of your own. If you already pay for DataForB2B, connecting it gives Kite another database to search, and those searches use your DataForB2B credits.