Agent types

Sema4.ai has two kinds of agent. You pick the type when you create one. Both are built the same way — a runbook, tools, and data — and differ only in how work reaches them and how you interact.

Conversational Agent

Interactive: you chat with it in a thread and it responds. Best for human-in-the-loop work — asking questions, exploring data, drafting, and one-off tasks where a person stays in the loop.

Worker Agent

Autonomous and queue-driven: it processes Work Items on its own, with no one watching. Work Items are created on a schedule or through the API (including file uploads), and each one moves through a tracked lifecycle — Pending → Executing → Completed / Error / Needs review — visible on the agent's dashboard.

Best for repeatable, high-volume back-office automation that needs operational control and an audit trail: document intake, reconciliations, and validations.

Choosing

• Use a Conversational Agent when a person drives the interaction.
• Use a Worker Agent when work should be picked up and processed automatically.

You're not locked in — an existing Conversational Agent can be copied as a Worker Agent to reuse its runbook and tools for unattended processing.