Tips, Tricks & Personalization We Use

Don't leave your AI setup blank. Load your context once so you never re-explain yourself again.

1. 01Open Settings → Custom Instructions (ChatGPT) or create a new Project (Claude).
2. 02Write 2–3 sentences: your role, your audience, your preferred format. Example: "I run AI workshops for adult professionals. No jargon. Bullets over paragraphs."
3. 03Save it. Every new chat now starts with that context already loaded. No intro needed.

Not all AI tools do the same job. Match the task to the tool before you start. It cuts your time in half.

1. 01Synthesizing your own documents? → NotebookLM. Upload files and ask questions about them.
2. 02Need live web information? → Perplexity. It searches in real time and cites every source.
3. 03Ongoing project that needs memory? → Claude or ChatGPT Project.
4. 04A task you repeat constantly? → Build a Gemini Gem. It knows your role before you say a word.

A Project is a persistent workspace. Your instructions, documents, and history all live there. No more starting from zero. starting from zero.

1. 01In Claude or ChatGPT, create a Project named after the work (e.g. "Workshop Content" or "Client Proposals").
2. 02Write a project instruction: your goal, your audience, your preferred tone. Be specific. It matters.
3. 03Upload 1–3 core documents: a past deck, a brief, a style guide. Now the AI has real context.
4. 04Every chat inside this project now inherits all of it automatically.

One giant prompt gets a mediocre answer. Break the task into steps and the quality jumps noticeably every time.

1. 01Step 1: Generate: "Write a first draft of [thing]."
2. 02Step 2: Critique: "What are the 3 weakest parts of that and why?"
3. 03Step 3: Fix: "Rewrite it with those weaknesses addressed. Keep it under 100 words."
4. 04Three focused prompts. Better result than one overstuffed prompt every time.

When a prompt fails, most people rephrase and guess again. Figure out what actually went wrong first.

1. 01Too generic? You didn't give enough context. Add your audience and goal explicitly.
2. 02Wrong tone? Say it: "Rewrite this like you're explaining to a curious colleague over coffee."
3. 03Wrong length? Be exact: "Under 80 words" or "At least 200 words."
4. 04Totally off? Ask: "What did you think I was asking for?" It usually reveals the miscommunication instantly.

AI sounds equally sure whether it's right or wrong. The advanced skill is knowing when to verify and how to make it surface its own weak spots.

1. 01After any output that matters, ask: "What are you least certain about in that response?"
2. 02Or try: "What would someone who disagrees with this say?"
3. 03Check any specific facts, stats, or names before using them publicly.
4. 04The AI drafts. You own the final product. That's the deal.