Reference material for using AI assistants and agents with clear prompts, appropriate boundaries, verification, and human accountability.
AI assistants can draft, summarize, analyze, transform, and retrieve information. Their output is probabilistic, so useful deployment combines a clear task, appropriate context, data controls, verification, and a person accountable for the result.
| Topic | Practical focus |
|---|---|
| AI Assistant Quick Reference | Prompt structure, output verification, sensitive data, and agent boundaries |
| AI Use Resources | Primary governance and technical resources |
1Define the task and intended user
2→ provide approved context and constraints
3→ request a usable output format
4→ verify accuracy, provenance, privacy, and policy fit
5→ revise or escalate before consequential use
Use AI to accelerate work, not to bypass responsibility. The higher the impact of an outcome—such as a security decision, legal communication, personnel action, medical content, or production change—the stronger the review, evidence, and approval should be.
| Term | Meaning |
|---|---|
| Generative AI | A system that produces new text, images, code, audio, or other content from learned patterns and input |
| Prompt | Instructions and context supplied to a model |
| Grounding | Supplying relevant, controlled source material so output can be tied to evidence |
| Hallucination | An output that is presented as plausible but is inaccurate, unsupported, or fabricated |
| Agent | A system that can pursue a goal through a sequence of steps, often using tools or external systems |
| Human-in-the-loop | A design in which a person reviews, approves, or can intervene in consequential decisions or actions |
The NIST AI Risk Management Framework provides voluntary guidance for incorporating trustworthiness considerations into AI design, development, use, and evaluation.