What makes an AI system agentic — autonomy, planning, tools, and memory — the perceive-plan-act-observe loop, and how it shifts the human from micromanager to supervisor.

A general reference on multimodal AI — how models represent text, images, audio, and video in a shared space, what they can do, how they are built, and their practical limits.

The metrics and methods for judging AI models — classification and regression metrics, robustness and efficiency, evaluation methodologies, and how generative systems differ.

The techniques past the basics — system prompts, structured prompts, reasoning and few-shot patterns, model parameters, iterative refinement, and evaluation — for predictable, controllable LLM output.

A technical overview of the Perplexity Search API: real-time, snippet-level web results for grounding LLMs, building agents, and feeding RAG pipelines.