How Kylon + Exa finds and verifies company and person context

Finding a company or person on the web is not the same as understanding them. Search results may surface pages that share the same words but miss the relevant business context. A sales, partnership, or recruiting team then has to read through sources, decide which facts are current, and explain where each claim came from.

On Kylon, Exa gives your agent an AI-powered way to enrich companies and people from public web information. No API key, no account, no configuration. The agent can use semantic matching to find relevant context, then return a concise profile with citations and confidence labels. That means the result can be reviewed, challenged, and used in the conversation rather than treated as an unexplained summary.

What you can do

Kylon's Exa integration is built around public company and person enrichment:

• Enrich companies with public information about what they do, their market, products, customers, positioning, and current context
• Enrich people with public professional context, roles, expertise, and relevant work history where available
• Use semantic search to find sources related by meaning, not just sources that repeat an exact keyword
• Retrieve citations so a reader can inspect the public pages behind a claim
• Apply confidence labels that distinguish strong, well-supported context from weaker signals
• Research an account list before outreach, a partnership discussion, or a strategic review
• Find comparable companies or relevant people based on business description and context
• Produce a compact brief that records what is known, what is inferred, and what still needs confirmation

Exa uses public web information. It should not be treated as an authoritative source for private employment data, personal contact information, or facts that require direct confirmation from the company or person.

How it actually works

A partnerships team may ask: “Give me a research brief on this company before our meeting. Explain its product, target customers, likely priorities, recent public developments, and the sources supporting each point.”

The agent calls Exa's company enrichment capability and returns a cited profile. Rather than a long list of links, the output can separate the company description, products, customers or market, recent developments, and areas that need validation. Each material point is linked to supporting public sources, and the agent can show a confidence level where the evidence is incomplete or inconsistent.

For a person, a recruiter might ask: “What public experience does this candidate have with developer tools and early-stage products? Cite the evidence and flag anything that looks uncertain.” The agent uses person enrichment to identify relevant public context and builds a focused summary. It does not turn scattered search snippets into certainty. It retains source references and says where the public record is thin.

Semantic matching is useful when the language used on the web differs from the language in your request. For example, a team could ask for companies that help finance teams automate close processes without using the exact phrase “financial close automation.” Exa can locate contextually relevant companies and sources, then the agent can filter the results by the criteria that actually matter, such as customer type, geography, or integration ecosystem.

Common use cases

Exa is most useful when you need a defensible understanding of a company or person, not merely a search result.

A traditional keyword search is still useful when you know the exact page or phrase you need. Exa is a better choice when the question is conceptual: “Who does this kind of work?”, “What is this company really focused on?”, or “Which sources support that conclusion?”

Getting started

Provide the company name, person name, or the type of organization you want to understand. Say what you need the research for. Examples include: “Prepare for a sales call,” “Assess partnership fit,” “Find comparable companies,” or “Summarize this person’s public experience in a specific area.”

Ask the agent to include citations and confidence labels when the brief will inform an important decision. It will use Exa to gather the relevant public context, distinguish evidence from inference, and return a reviewable result directly in the Kylon channel.