TPEE Toptronic® Prompt Engineering Editor

60. Common Beginner Mistakes and How to Fix Them

Recognize and avoid the most common errors new users make.

By Jacques Botte, founder of Toptronic®. Last updated 19 September 2026.

The lesson

Mistake 1 - Vague instructions: Writing 'fix this' without explaining what 'fix' means or what success looks like. Fix: Be specific about the problem and desired outcome.

Mistake 2 - Missing context: Asking for code help without showing the error message or file structure. Fix: Always include relevant error text, file paths, and constraints.

Mistake 3 - Wrong persona: Using a generic persona for specialized work. Fix: Match the persona to the task - use 'Rust Expert' for Rust code, 'Business Analyst' for reports.

Mistake 4 - No format specified: Letting the AI guess the output format. Fix: Explicitly state 'Give me a numbered list' or 'Format as a Markdown table'.

Mistake 5 - Trusting blindly: Using AI output for critical decisions without review. Fix: Always verify important work, especially code, legal, medical, or financial content.

Mistake 6 - Not using presets: Recreating the same prompt structure repeatedly. Fix: Save good prompts as presets for reuse.

A retail shop assistant replaces a vague instruction with a specific request, then checks every price claim the model returns before using it.

A junior aged care worker adds the resident's care context after noticing that the model had guessed details she left out of the prompt.

An apprentice plumber swaps a generic assistant persona for an experienced trades writer when drafting a job explanation for a customer.

A new veterinary receptionist specifies the exact output format she needs after a first attempt returned one long unusable paragraph.

A first-year mining site clerk pastes the exact error text from a spreadsheet export instead of describing it, so the answer addresses the real problem.

A graduate accountant spots a figure the model invented, then grounds the prompt with the trial balance before reusing the answer.

A new hospital porter saves a prompt that worked well as a preset instead of retyping the same structure every shift.

A junior hospitality bookings clerk fixes a wrong-persona mistake by naming a guest-services writer, then checks each refund line against the policy manual.

A trainee laboratory assistant learns to name the instrument and the units in the prompt after a first attempt returned unrelated advice.

A new insurance call-centre operator asks for a numbered checklist after her first reply came back as unstructured prose.

Check yourself

Question 1: What is 'Mistake 1' that beginners often make?
  1. Writing too much context
  2. Vague instructions like 'fix this' without specifics — correct
  3. Using too many technical specs
  4. Saving too many presets

Answer: Vague instructions like 'fix this' without specifics

Vague instructions like 'fix this' leave the AI guessing. Always specify what needs fixing and what success looks like.

Question 2: Why is using the wrong persona a mistake?
  1. It has no effect
  2. The AI may not have the right expertise for your specific task — correct
  3. It makes the text colorful
  4. It is required by law

Answer: The AI may not have the right expertise for your specific task

Different personas represent different expertise areas. Using 'Rust Expert' for Rust code gives better results than a generic persona.

Question 3: Why should you not blindly trust AI output?
  1. AI is always wrong
  2. AI can make mistakes; important work needs human verification — correct
  3. It is illegal
  4. AI output is encrypted

Answer: AI can make mistakes; important work needs human verification

AI can hallucinate or make errors. Always verify important work, especially for code, legal, medical, financial, or safety-critical decisions.

← Previous lesson · All 91 lessons · Next lesson →

The full course — 91 lessons and 273 quiz questions — ships inside the app. Get TPEE to study it offline.