AI & Agents.

AI workflows create more than a final answer. They retrieve context, generate interpretations, propose tools, and produce artifacts that may require a person to review them. This collection gives those stages separate identities. Begin with the AI record model for jobs, attempts, validation, and approval. Continue into MCP and LLM context or agent traces when the workflow crosses tool boundaries. The media guides then apply the same provenance and review principles to audio and video. Throughout the collection, examples are architecture proposals rather than claims about a hosted service. Use them to decide what your application can observe, what it needs to retain, and which results should remain uncertain until evidence or a reviewer resolves them.