AI glossary

Terms as Aarohii uses them on this site. Longer explainers live under Learn.

AI agent: Software that plans steps and may call tools. Full definition
AI automation: Models inside a repeatable process. Full definition
AI governance: Owners, policy, review, shutdown. Full definition
Embedding: A numeric representation of text used for similarity search in RAG.
Evaluation: Repeatable tests of model or retrieval quality on tasks you care about. Full definition
Fine-tuning: Further training of a model on your examples. Often later than RAG. Full definition
Generative AI: Models that create new content. Full definition
Hallucination: Plausible output not supported by sources or facts.
Human-in-the-loop: A required human approval before an irreversible action.
LLM: Large language model. Full definition
Observability: Logs, traces, and metrics so you can see what the system did. Full definition
Prompt engineering: Designing and versioning instructions to the model.
RAG: Retrieval-augmented generation. Full definition
Tool calling: The model selecting an API or function to run.
Vector database: Storage for embeddings used at retrieval time. Full definition

If the problem maps to work we actually ship, we will say so in 20 minutes.