AI agent
also: LLM agent · tool-using AI · agentic AI
A system in which a language model plans and carries out multi-step tasks by calling tools, such as searches, databases or device APIs, to reach a goal.
Instead of answering a single question, an agent works in a loop: decide the next step, call a tool (search documents, query a database, run a calculation, read telemetry, open a ticket), look at the result, and continue until the task is done or a human is needed. In hardware settings, agents can triage field alarms, draft test reports from logs, check part availability, or guide technicians through diagnostics.
Because agents act, safety design matters more than for chat: limit which tools they can use, require human approval for consequential actions (anything that changes equipment state, spends money or contacts customers), log every step, and evaluate on realistic tasks. Frameworks such as the NIST AI Risk Management Framework give structure to these decisions.
Founder takeaway
Give agents the narrowest tool access that does the job, and keep human approval for any action that changes physical equipment or makes commitments.
engineer review pending