Getting Started
With AI the
Right Way

If you tried ChatGPT in 2024 or early 2025, asked it to write an email, got back something that sounded like a robot ate a thesaurus, and decided AI was overhyped — you were right. At the time.

But here’s what you missed: AI in 2026 is not the same thing you tried. The models are dramatically better. The tools are completely different. And the way people actually use AI has changed more than the technology itself. The gap between someone who tried AI once and someone who uses it daily is now like comparing a 2005 flip phone to an iPhone 16. Same category. Completely different experience.

Here are the three things that were probably wrong when you tried it:

1. You gave it zero context. You typed something like “write me an email” and got back a generic, corporate-sounding mess. That’s because AI without context is like asking a stranger to write an email on your behalf — they don’t know who you are, how you write, who you’re writing to, or what you want to say. Of course it was bad.

2. You used it for the wrong things. Most people start by asking AI to write something from scratch. That’s actually one of the hardest things for AI to do well. The things AI is incredible at — analyzing data, summarizing long documents, comparing options, finding patterns, organizing chaos, brainstorming, researching — most people never try.

3. You used the wrong tool. ChatGPT in 2024 was the first wave. The tools available now — Claude, Perplexity, NotebookLM, Wispr Flow — are purpose-built for different things and they’re dramatically better at what they do. Using 2024 ChatGPT to judge AI in 2026 is like using Internet Explorer to judge the internet.

Tool 1

Claude (claude.ai) — your AI brain for everything. This is the one you use every day. Think, write, plan, analyze, brainstorm, research, organize — Claude does it all. The free plan gives you a generous amount of usage. The Pro plan ($20/month) gives you unlimited access, the ability to connect your Gmail and Calendar, and features like Projects (where Claude remembers context across conversations). Start here. This is the big one.

Tool 2

Perplexity (perplexity.ai) — what Google should be. When you need to research something, look something up, compare options, or find current information — use Perplexity instead of Google. It searches the web, reads the results for you, and gives you a clean answer with sources. No ads. No scrolling through 10 blue links. Ask a question, get an answer with citations. Free to use.

Tool 3

NotebookLM (notebooklm.google.com) — the best free AI tool nobody talks about. Upload any documents — PDFs, articles, notes, textbooks, reports — and NotebookLM reads them and lets you ask questions about them. It can turn a 50-page report into a 2-minute summary, create study guides, build FAQs, and even generate an AI-hosted podcast from your documents. Completely free. Made by Google.

Tool 4

Wispr Flow (wispr.com) — talk instead of type. Wispr Flow runs in the background on your computer. Instead of typing, you just talk, and it turns your voice into clean, edited text — anywhere you type. Emails, Slack messages, documents, forms, text fields. It removes filler words, fixes grammar, and formats everything. It’s genuinely faster than typing. Free plan available.

Go to claude.ai and create a free account. You can also download the desktop app (Mac or Windows) or the mobile app. Once you’re in, you’ll see a text box. That’s where you talk to Claude. But how you talk to it determines whether you get genius-level output or generic garbage.

Here are 7 prompting rules that change everything:

Rule 1: Tell Claude who it is. Start your prompt with a role. “You are a senior marketing strategist with 15 years of experience in B2B SaaS.” This alone makes the output 10x better because it gives Claude a perspective to write from instead of trying to be everything to everyone.

Rule 2: Give context before asking for anything. Don’t just say “write me an email.” Say “I’m a project manager at a tech startup. I need to email my team about a deadline change. The original deadline was Friday, the new deadline is next Wednesday. The team is stressed and I want the tone to be calm and reassuring. I usually write in a direct but friendly way. Write the email.” That’s 10 seconds of extra typing and a completely different result.

Rule 3: Tell Claude what you DON’T want. “Don’t use corporate buzzwords. Don’t start with ‘I hope this email finds you well.’ Don’t use bullet points. Keep it under 150 words.” Constraints make AI output dramatically better. Without them, Claude defaults to safe, generic, long-winded responses.

Rule 4: Paste in real content. Claude is incredible when you give it something to work with. Paste in a meeting transcript and ask for action items. Paste in a job posting and ask how your resume matches. Paste in a contract and ask it to explain every clause in plain English. Paste in 3 months of sales data and ask for patterns. AI analyzing your real content is 100x more useful than AI generating content from nothing.

Rule 5: Have a conversation, not a command. Your first prompt doesn’t have to be perfect. Say something, see what Claude gives you, then refine. “That’s too formal, make it sound more like how I’d actually talk.” “Good, but add more specific data.” “Cut this in half.” The best AI users go back and forth 3–5 times. The worst AI users send one prompt and judge the first response.

Rule 6: Use Claude for thinking, not just writing. Ask Claude to poke holes in your business plan. Ask it to play devil’s advocate on a decision you’re about to make. Ask it to list 10 things that could go wrong with your project. Ask it to compare two options side by side. The most powerful use of AI isn’t writing — it’s thinking alongside you.

Rule 7: Set up your profile. Go to Settings → Profile and tell Claude about yourself. Your name, your job, how you write, what you care about. This is like introducing yourself to that new colleague. Do it once and every conversation after that starts on better footing.

Do This — In This Order