Jedify launches Data Agents for plain-language AI analysis

Jedify launched Data Agents, now available through its Agents Hub, for business teams to configure analysts by describing their desired work in plain language. The company says the agents can monitor metrics, investigate changes and run recurring analyses without requiring users to write code or assemble workflows.

The product is designed to address questions that teams did not anticipate when building dashboards. Users can define an outcome conversationally, refine the agent and direct it to investigate a new question or respond when a specified business condition occurs. Jedify is pitching the system to data teams and business functions including finance, revenue operations, customer success, product and sales.

Jedify says its context graph gives each agent organization-specific definitions before a query runs. The graph maps schemas and joins, metric definitions, relationships and business rules, so the agent does not have to infer those meanings from raw tables each time. The company claims this approach uses fewer tokens than agents that explore raw schemas at query time. The launch materials do not quantify that claim, however, and the supplied sources do not independently test it.

Agents can run on demand, on a schedule or when a user-defined condition is met. Jedify also says each agent inherits the requesting user’s existing data permissions, an approach intended to prevent it from returning rows, columns or metrics that person is not allowed to see. The release does not describe the technical enforcement or audit controls behind that permission inheritance.

Through the Model Context Protocol, also used to connect external agents with productivity software, a Data Agent can be invoked from Claude, ChatGPT, Slack or another compatible agent platform. Results can also be delivered through Slack or email. Jedify says organizations can start using Data Agents for free, although the launch materials do not provide paid pricing or free-tier limits.