← AI 101
Beginner training

AI For Dummies

A practical Claude starter module for employees. Learn what Claude is, when to use it, how to ask for useful work, and what to double-check before you trust the answer.

20 minutes No AI experience needed Includes practice tasks

What employees should know after this lesson

01

Explain Claude simply

Claude is an AI assistant you can talk to. It can draft, summarize, compare, analyze, and help plan work.

02

Choose the right task

Use Claude for language-heavy, document-heavy, and thinking-heavy work. Use other tools when a job needs live facts, exact math, or image generation.

03

Give better instructions

Useful answers come from clear context, examples, constraints, and follow-up. Treat prompting like delegating to a capable teammate.

How Claude works, without the engineering lecture

A

It predicts language

Claude builds responses by predicting useful next pieces of text from your prompt and the patterns it learned during training. It can feel like thinking, but it still needs human direction.

It tries to be helpful

Because it is optimized to cooperate, it may agree too easily. Do not confuse a confident tone with correctness.

It uses tokens

Claude reads and writes text in small chunks called tokens. Long files, long chats, and long answers all consume that working space.

Trainer note: The simplest mental model is not “magic brain.” It is “very fast pattern matcher that gets much better when you provide context.”

Where employees will meet Claude

Browser Claude

Best for quick questions, drafts, summaries, and low-risk practice.

Start here

Desktop Claude

Useful when employees need a richer workspace and access to local files or folders.

Next step

Cowork or agent-style work

Best for longer jobs where Claude can inspect material, plan steps, ask questions, and produce deliverables.

Advanced

Five rules for getting useful answers

1

Be specific

Name the audience, goal, tone, format, deadline, and constraints.

2

Give examples

Paste a sample of work you like. Claude learns faster from examples than from abstract preferences.

3

Say what good looks like

Tell it what to produce, not only what to avoid. Clear positive direction beats vague warnings.

4

Iterate in short rounds

Start with a compact prompt, read the answer, then ask for changes. The conversation is part of the work.

5

Restart when the chat gets messy

If the answer drifts, open a clean chat and paste only the important context.

Weak prompt

Write an email about the meeting.

Work-ready prompt

Write a friendly but direct follow-up email to a client who missed our Tuesday call. Include the new booking link, mention that we need their decision by Friday, and keep it under 120 words.

What Claude is especially useful for

Writing and rewriting

Draft emails, simplify internal updates, turn notes into polished summaries, or adapt writing to a known voice.

Document digestion

Summarize long PDFs, extract action items from meeting notes, compare versions, and ask for page references when possible.

Thinking partner work

Pressure-test a plan, create options, identify risks, and ask it to argue both sides before you decide.

File-based workflows

For approved workspaces, Claude can help organize folders, draft outputs, or turn raw material into structured deliverables.

What employees must verify

Employee rule: Claude can help create and critique work. The employee is still responsible for judgment, verification, and final approval.

The first-week challenge

Pick three exercises. Save any prompt that produces a genuinely useful result.

  1. Paste a recent email and ask Claude to rewrite it in the same voice, but clearer.
  2. Upload or paste meeting notes and ask for decisions, owners, deadlines, and open questions.
  3. Ask Claude to summarize a long document into one page with a verification checklist.
  4. Give Claude a rough plan and ask it to find risks, missing steps, and assumptions.
  5. Paste a messy customer thread and ask for the next three actions.
  6. Run the same task in another AI tool and compare which answer is more useful.

Completion standard

An employee has completed this module when they can explain one good use case, one risky use case, and write a prompt with context, constraints, and a desired output format.