Troiana AI Hub

Get real work out of AI.

Free courses, tool guides and quizzes for people who use AI to get something done. Start from the job in front of you, or from the tool you already have open.

5 courses266 techniques187 tool guides40 quizzes

The complete guide

Vibecoding

Build real software by describing it, and keep your users' data safe when you launch.

7 modules · 29 techniques
Replit's Agent building a workshop website, with the chat on the left and a live preview on the right.
Start from the job

What are you trying to do?

Each one goes straight to the lesson or guide for that job. These are starting points based on what each tool is built for, not benchmark results.

How to use AI tools well

The same four habits work in every tool.

  1. 1

    Describe the finished result

    Say what done looks like, who it's for, and what format you want. A one-line request makes the model guess all three.

  2. 2

    Give it what it can't guess

    Paste the source, the example or the file. A weak answer often comes from missing context, not from a missing magic phrase.

  3. 3

    Match the model to the task

    Use a fast model for quick work and a reasoning model when a wrong answer would cost you. Stronger is not always better.

  4. 4

    Check the part that matters

    A confident answer can still be wrong. Find the number, quote or assumption the result depends on, and check that against a source.

Your prompt, ready to paste into any assistant
I need: [the result, in one sentence]
It's for: [who will read or use it]
What you should know: [facts, files or examples you can't guess]
Limits: [length, tone, tools, anything to avoid]
Done means: [how I'll judge it]

If something important is missing, ask me up to three questions before you start.
When you finish, list any claims I should verify.
New to all of this? Start with what a large language model is
Guides

Every tool, explained.

Setup, pricing, limits and lesser-known tips for 187 AI tools, written from each vendor's docs.

Browse the guides
Quiz

Try one right now.

40 quizzes with instant answers and no sign-up.

Your app has a working login screen. What has that proved so far?

See all 40 quizzes
Library

See what it made. Copy what made it.

Real AI results, each with the prompt or skill behind it.

Open the Library
Frequently asked
Which AI tool should I use?

Start from the job, not the brand. For general questions and writing, ChatGPT, Claude and Gemini are all general assistants, and each has a free plan, so try your real task in two of them and compare. To change code inside a project, use a coding tool such as Cursor or Claude Code. For an answer with sources you can open, try Perplexity. The table of jobs on this page links to the right lesson or guide for each one.

How do I get better results from any AI tool?

Describe the finished result, give the model the context it can't guess, match the model to the task, and check the part of the answer that matters. Those four habits carry across every tool and do more than any single prompt. The prompt template on this page puts them into a form you can copy.

Do I need to learn prompt engineering?

You need the basics: clear context, a defined result, and an example of what good looks like. You don't need a collection of tricks. Choosing the right tool, managing what the model can see, and checking its output usually matter more.

Is this for beginners or advanced users?

Both. Each course starts with first steps and moves up through real workflows and advanced techniques, and every technique is tagged by level. If you already use a tool daily, take its quiz first and let your wrong answers show you where to go.

Is it free?

Every course, guide and quiz here is free, with no sign-up. The tools themselves are priced by their vendors. Plans change often, so the tool guides show the date they were checked and link to the vendor's own pages.

Who writes this, and how is it checked?

Troiana, a design and development studio in Prishtina. Tool guides are written from each vendor's official docs, with sources linked on the page. They report what the docs say, and they don't claim hands-on benchmarks.