Tool use
When an AI model does more than produce text by calling on outside tools, such as searching a database, looking up a record, running a calculation, or sending a request to another system. Tool use is what lets a model act in the real world rather than just describe it, and it is the core of how agents get things done. For a healthcare audience this is where capability and risk both rise sharply: the moment a model can reach into a real system, what it is allowed to touch, and what it is blocked from touching, becomes a safety question, not just a feature.
An assistant that drafts replies to patient portal messages can look up the chart and the clinic schedule, and the vendor offers an upgrade that lets it book and cancel appointments directly. Reading a schedule and changing one carry different risks: a wrong draft gets caught by the nurse who reviews it, while a wrongly canceled follow-up may go unnoticed until the patient misses care. Ask for a list of every tool the assistant can call, mark which ones only read and which ones change something, and require a staff member's approval for each change.
Terms like this come up in real clinical scenarios across the HelloAI courses: bite-sized modules with verifiable certificates. An account takes one minute, no password needed.
Sign in →