You are here. if you’re new. Otherwise:
- MODELS — The AI engines themselves. One at a time, directly. Models
- TOOLS — Apps built on top of an AI to do a specific job. Tools
- ORCHESTRATION & AGENTS — Systems that choose or coordinate which AI answers you. Orchestration
- TERMS — The words you keep hearing, in plain language. Terms
Have a specific job to do? Go to Tools. Wondering why your AI seems to have gotten worse lately? That’s Orchestration.
THE SHORT ANSWER is that for low-stakes questions, most free models are often good enough. But you won’t really understand what all the fuss is about.
To get a better understanding you need to try one of the best models (like Claude Opus, Fable, ChatGPT 5.6 Sol, Kimi K3 or Z.ai GLM 5.2) and turn the thinking up. Ethan Mollick has a good guide here (July 2026): the exact ranking change weekly but that is mostly for super-specific and difficult work. These models have lower error rates on hard problems. They also cost money, but a paid model will teach you more in a month than forever on free ones. Start with the cheapest tier ($20). The expensive tiers mostly buy more hours of AI labor, not a smarter AI. Pay for a month and give it hard things to do. Pay for the higher tier when you are running long complicated or agentic tasks, not because you expect better answers.
YOU MAY NOT GET TO CHOOSE: Most advice about which AI to buy assumes you are buying. If your campus has an enterprise license, that is your model. The best you do is know what it is good and bad at. Check also whether you are permitted to put student work into your (paid) personal account. Probably not.
TRY AN AGENT on your desktop. Agents are scary, but you can isolate their access to a specific website or folder on your desktop. I find Claude Cowork very careful, and the defaults are to ask permission. Give an agent one real task from your actual work (like asking it to fix the menus on your website). Then look hard at what comes back and ask for changes, the way you would with a person. This will change your understanding of how AI is going to get built into all human work.
Where to go next
- Models is the place to start if you want to know what models exist (MANY more than you probably thought you knew!) and how the engines/brains differ.
- Tools is where to go with a specific job in hand — a literature review, a set of slides, a grading pile. Tools are built on top of models.
- Orchestration & Agents explains who is choosing which AI answers you, including inside your LMS, and how the newer routers and harnesses work. You will also find how consolidators, routers and tools like Perplexity and BoodleBox fit in.
- Under the Hood has the terms and the plumbing, for when something behaves in a way you did not expect.