Learntova.
Learn by doing

A small space for big ideas.

Build a clear prompt, then try it in an AI tool of your choice.

Guided practice

Build a better prompt

4 simple ingredients
Your prompt, put together
Your prompt will take shape here as you fill in the four ingredients.

Make it your own.
A useful prompt gives just enough detail for the task. More words don’t always mean a better answer.

This lab builds and checks your instructions using a simple checklist. It does not send prompts to an AI model. Use fictional data, and verify any output you get from an external AI tool.

Learn skills that travel.

You do not need every AI tool. Start with one, then compare when you have a reason.

ChatGPT

Try a conversation, draft a message, or revise a plan.

Open ChatGPT

Claude

Try the same writing prompt and compare the result.

Open Claude

Gemini

Practise explaining, drafting, and asking follow-up questions.

Open Gemini

Canva

Turn a reviewed message into a visual communication.

Open Canva

These are external services with separate accounts, terms, and usage limits. Learntova payment does not include their subscriptions.

Compare two answers

  1. Use the same fictional prompt in two tools.
  2. Check the facts, missing information, and whether each followed your instructions.
  3. Write down which result better fits your purpose and why.
  4. Keep the useful parts and revise the rest yourself.

A longer or more confident response is not automatically a better response.

A little less jargon

Plain English for the words this course uses.

Prompt

The instructions or question you give an AI tool.

Context

Background information that helps the tool understand your task.

Hallucination

An AI answer that sounds plausible but contains invented or unsupported information.

Iteration

Making a change, checking the result, and improving it again.

Verification

Checking a claim against a reliable original source.

Workflow

A sequence of tasks with inputs, checks, and a useful result.

Model

The AI system that generates answers from patterns it learned during training.

Training

The process in which a model learns patterns from large amounts of text or other data, mostly before you use it.

Template

Saved instructions you reuse, with placeholders for the details that change.

Placeholder

A marked gap in a template, such as [meeting notes], that you replace each time.

Grounding

Giving the tool reference material and asking it to answer only from that material.

RAG (retrieval-augmented generation)

A system that finds relevant reference passages automatically and gives them to the model when it answers.

Structured output

An answer in a fixed shape, such as a table or JSON, so each field can be checked.

JSON

A common text format for structured data, made of named fields and their values.

Null

A value that means “not known” or “not supplied”, used instead of guessing.

Evaluation set

A fixed group of test cases, with expected results, used to check whether a change helps.

Prompt injection

Text that tries to change an AI tool’s instructions, typed directly or hidden in a document it reads.

Permission

What a tool is allowed to do, such as read files or send messages, set outside the prompt.