Whozta Product · 3 min read
Introducing Whozta: the people-search layer for agents
Why agent workflows need a reliable resolution layer between a fuzzy request and the person an agent can act on.
The question behind the product
Agents are getting good at understanding intent. The hard part is still turning a request like “find the person who owns partnerships at this company” into a useful result that can support a real next action.
That request contains ambiguity: the company may have several partnership teams, titles may differ, and public information can be out of date. A name without context is not enough for an agent to decide whether to act or ask a follow-up.
Resolution needs evidence
Whozta is designed as a resolution layer: start with a company and a role, then return a candidate together with source context, contact details, and a confidence signal. Those pieces let the downstream agent choose between acting, asking a clarifying question, or continuing its search.
The goal is not to hide uncertainty behind a single score. It is to make the reasoning surface usable so a developer can set the right review threshold for outbound, recruiting, CRM enrichment, or support routing.
Designed for agent interfaces
The same resolution workflow should be available where an agent already works. MCP makes the capability available as a native tool, while a REST API makes it possible to integrate the same behavior into a custom application or background job.
Keeping the interfaces close to the underlying model also makes it easier to observe failures, refresh stale information, and preserve provenance as a result moves through an agent workflow.
What we are learning
The hard problems are not limited to finding a name. They include matching roles that are expressed differently, separating current information from stale pages, and presenting enough evidence for a person to review the result responsibly.
Future posts will cover those problems in more detail, along with the practical interfaces and safeguards that make people search useful inside production agent systems.