Day: 126 Level: Intermediate Edition: Claude Code Category: Costs & Usage AI Learning Path: Real-World Usage Tokens are the basic units that Claude Code uses to measure and bill your usage. Understanding tokens is fundamental to understanding Claude Code costs — every prompt you send and response you receive consumes tokens, which directly affects your bill. AI Terms in Series: Tokens → Models → Subscription Level → API Use → Extra Usage 🪄 Simple Definition: The units that Claude Code uses to measure and bill your usage.→ “How Claude Code counts what you use.” 🌟 Expanded Definition: A token is roughly equivalent to a word or part of a word. When you interact with Claude Code: Input Tokens (what you send): • Your prompts and instructions • Code you ask Claude to review • Context from your project files • Typically cheaper than output tokens Output Tokens (what Claude sends back): • Claude’s responses and explanations • Code that Claude writes • Usually costs more than input tokens How Tokens Add Up A typical interaction might use: • Simple prompt: 50-200 tokens • Code generation request: 200-1,000 tokens • Complex project with context: 2,000-10,000+ tokens Token Costs Vary by Model: • Claude Haiku: Cheapest per token • Claude Sonnet: Mid-range pricing • Claude Opus: Most expensive per token ⚖️ However, you can’t choose a model based solely on token price — you need to consider the model’s capability for your specific task. A cheaper model that can’t handle complex requests might cost more in the long run through multiple attempts or poor results. Understanding tokens helps you predict and control costs. Tokens are essential for: • Estimating costs before large requests • Understanding why complex prompts cost more • Optimizing your prompts for efficiency • Managing your Claude Code budget • Choosing between models based on value ⚡ In Action: “You ask Claude Code to build an API integration. Your prompt uses 300 input tokens, Claude’s response generates 2,000 output tokens. At Sonnet 4.6 pricing ($3/$15 per million tokens), this interaction costs about $0.031 — understanding tokens helps you budget for bigger projects.”

Ever struggle to get the right facial expression for a YouTube thumbnail or LinkedIn image? I just created an amazing array of headshots using Gemini + Canva with different emotions while staying close to my actual likeness and personal brand. A few things I liked about it: - Different expressions gave me more thumbnail options - Asking for a little more torso made the images easier to use in layouts - Pulling the final image into Canva made the design side fast and flexible This is one of those small use cases that can be surprisingly practical if you create content regularly. Sometimes the hardest part of a thumbnail is not the design. It is finding the right look to match the message. Have you tried using AI for headshots, expressions, or thumbnail concepts yet? View Now (2.5 Minutes - Yes that is it) https://youtu.be/LbYtViAilGQ?si=ghFk8mgVWFVkVe9O

I ran a live session walking through a real life example of using NotebookLM to turn raw source documents into structured insights, summaries, and visual outputs. If you want to understand: - what NotebookLM actually is - how it differs from LLM's like ChatGPT - what you can practically do with it This is a good place to start. One of the key things that makes NotebookLM different is that it works from the sources you give it — documents, notes, links, and files — rather than pulling from the open internet. In this session, I walk through: - what NotebookLM is - how it works with your own data - how to add and manage source documents - how to ask questions and get grounded answers - how to turn notes into new sources - how to create summaries, quizzes, and infographics

If you’ve been hearing about NotebookLM but haven’t really used it yet, or are interested in seeing how I use it - this session is a good place to start. We’ll walk through the fundamentals—no assumptions, no prior setup needed. 🔍 What we’ll cover: • What NotebookLM actually is (and what it’s not) • How to add and structure source material • Turning messy information into usable notes • Creating summaries, FAQs, and simple workflows • Where this becomes useful in real day-to-day work This session isn’t about advanced builds - it’s about getting comfortable with how the tool works so you can start using it with your own content. 🔍 Who this is for: • AI Curious — you’re learning what’s possible • AI Enthusiasts — you’re using tools but want better structure • Anyone sitting on content they haven’t fully used We’ll keep it practical and move at a pace where you can follow along. 📅 Live Session April 15, 2026 1pm EST

We just wrapped Session 2 of our live build series and it was a great one! Starting from absolute scratch, we built a fully functional Task List app in under 10 minutes of total “cook time” using nothing but natural language prompts in Claude Code. No traditional coding. Just clear instructions and AI doing the heavy lifting. Here is everything we built in one session: ✅ Add, edit, and delete tasks ✅ Priority levels with color coding ✅ Due dates, categories, and status tracking ✅ Notes field per task ✅ Live progress bar ✅ Search, filter, and sort ✅ Collapsible completed tasks section ✅ Real SQLite database ✅ Runs locally on your machine The best part, while I was feeding Claude Code with prompts to build an enhanced Task List, several conversations broken out between @Matthew Sutherland @Bruce Kaufmann and I discussing many planning and building tips that only experienced builders could offer. The biggest lesson from today? Prompting is the new coding. If you can describe what you want clearly and in phases, you can build real working software — even if you have never written a line of code in your life. The YouTube video is now live. Go check it out and follow along. All the prompts we used are included. 👉 https://youtu.be/oJwNzUDzsIA?si=peazUxtSTJ8Y4rKA Session 3 is coming up on Monday. #ClaudeCode #VibeCoding #AIBitsAndPieces