| Aspect | Health AI agent | Chatbot |
|---|---|---|
| Unit of work | A job with a start and a finish. | A conversation. |
| Starts on | A message, a store event, an API call or your team. | A message. |
| Time | Waits days for a scan or a call, then carries on. | Ends when the chat ends. |
| Systems | Reads and acts through connectors, inside set permissions. | Text answers. |
| Safety | Red-flag check on every message, and a person one step away. | Depends on the build. |
| Record | Every trigger, step, tool call and error. | A chat log. |
Which to pick.
Pick an agent when the work has steps, needs your systems, or must be on record.
A chatbot can be enough for a fixed set of questions with no follow-up.
Questions
Health agent vs chatbot: what is the difference?
A chatbot answers messages; a health AI agent owns a job end to end, keeps state for days, waits, retries, uses your systems and records every step.
Is a health agent just a chatbot with a model behind it?
No. A model that talks well is not an agent. An agent owns a job end to end, and the job decides what it does next.
Can a health agent also chat?
Yes. Agent loops hold conversations, and can start a workflow or a call mid-conversation.
Does either one diagnose?
Ollie’s agents do not. They give wellness guidance and hand red flags to a person.