Paid courses on neural networks often rehash what model developers share for free. Below are eight primary sources: where to learn, where to get prompts, and how to extend agents.

Learning

  • LearnAIGoWhere — a collection of free courses from OpenAI, Google, Microsoft, and other companies that develop models themselves.
  • Hugging Face Learn — courses on machine learning alongside the largest catalog of open models. For a narrow task, you can often find a specialized model there that works better than a general-purpose chatbot.
  • Google AI Tips — 40 short tips on working with Google AI tools.

Prompts

  • PromptsToCheck — a library of templates for text, images, analytics, and in-depth research.
  • Prompt Mania — prompts for photorealistic images with adjustable lighting, lens, and environment.
  • Alex Prompter thread — eight prompt crafting techniques compiled from Anthropic and OpenAI engineers' materials.

Agents and Code

  • Superpowers — an open library of ready-made workflows for programmer agents: planning, debugging, result verification.
  • SkillsMP — a catalog of tens of thousands of skills for Claude Code and Codex. A skill hints to the agent how to work with a specific library or type of task.

The Right Order

  1. Take one introductory course to understand how models work and where they fail.
  2. Take ready-made prompts for your task and figure out why they're written that way.
  3. When chat becomes limiting, switch to agents and skills.

Skills and workflows from catalogs are third-party instructions and code. Before installing, read what they actually do.