New: Roadmaps ordered paths through our cheat sheets and flashcards, so you always know what to study next.
Explore themSee what's new on GitHubFrom your first well-worded prompt to choosing the right AI tool for any task.
A 12-step learning path. Follow it in order, or jump to what you need.
This path is for anyone who talks to ChatGPT, Claude, or Perplexity and wants noticeably better answers out of them, no coding background needed. Expect about 4 to 6 weeks at a few hours a week, moving from what a prompt actually does through the reasoning tricks and context habits that make AI tools reliable, then into using ChatGPT, Claude, and Perplexity with real skill. It does not cover building AI applications, retrieval pipelines, or fine-tuning a model on your own data (that is Generative AI & LLM Engineering's territory), and it does not cover AI coding agents for software development. By the end you can write a prompt that gets what you want on the first try, choose the right AI tool for a research task versus a first draft versus an image, and catch a confident-sounding wrong answer before you trust it.
No prior experience needed. Start from zero.
Open this and you'll recognize every AI tool you already use, chat models, image generators, code assistants, as branches of the same idea: predicting the next likely piece of content.
Now put that picture to work: the exact wording, structure, and role you give a prompt is the single biggest lever over what step 1's models hand back.
You can shape a chatbot's answer with wording alone, no special tools needed. Next up: examples and step-by-step reasoning that push accuracy even further.
Finish this section to unlock.
+100 XP
Stuck getting a vague answer from step 2's best-worded prompt? Show the model two or three examples of the output you want and watch it lock onto your pattern.
Step 3's examples still trip on multi-step problems: ask the model to show its work first and watch its accuracy on math, logic, and planning jump.
This is where it clicks: neither step 3's examples nor step 4's reasoning steps taught the model anything permanent, it just read them fresh in your prompt, and most learners circle back to this page more than once before that fully settles.
You can supply examples or ask for step-by-step reasoning to fix a weak answer, and a few due flashcards on this chapter keep those ideas fresh while you build on them next. Next up: putting all of it to work inside the tools you already use every day.
Finish this section to unlock.
+100 XP
Zoom out from a single message to the whole conversation: everything from step 5 lives inside a context window with a limit, and managing what stays in it or gets dropped is what keeps a long chat useful instead of confused.
Put steps 2 through 6 to work in the tool most people open first: custom instructions, memory, and projects turn ChatGPT from a one-off Q&A box into an assistant that actually remembers your context.
Run the same playbook against Claude and notice where it diverges from step 7: Projects and long-context handling reward giving it more upfront context rather than shorter, punchier prompts.
Neither step 7 nor step 8 promises its facts are current: for research where you need cited, checkable answers instead of a fluent guess, Perplexity's search-then-synthesize approach is the safer default.
You can lean on ChatGPT, Claude, or Perplexity for the right job, keep a long conversation from losing the thread, and check a claim against its cited sources. Next up: creating images and picking the right tool on purpose instead of by habit.
Finish this section to unlock.
+100 XP
Take this if you're curious what's actually happening when you drop a photo or PDF into step 7 or step 8's chat box: a shared embedding space is why the model can describe an image at all.
Aim here if you're the one generating images, not just consuming them: word choice, negative prompts, and style keywords control output far more precisely than steps 2 through 6 alone.
Bring it together: match steps 7 through 9's tools, plus step 11's image generators, to the actual task in front of you instead of defaulting to whichever app you opened last.
You can prompt with intent, verify what comes back, and pick the right tool for a research task, a first draft, or an image. That's what the Everyday AI Practitioner badge marks: not just knowing the tools exist, but knowing how to actually use them.
Finish this section to unlock.
+100 XP
Finish every required step, at least 70% of them genuinely done (not skipped), to earn this badge and 500 XP.