The big idea
Most people treat AI like a slightly smarter Google search — type a question, get an answer, close the tab. That captures a tiny fraction of what these tools can actually do in 2026. Getting to the rest doesn't require becoming a programmer; it takes a better mental model, a small set of skills, and the habit of building systems instead of just having conversations.
Deep dive
The gap between "using AI" and "working with AI"
Most people's relationship with AI looks like this: open ChatGPT a few times a week, type a question the way you'd type it into Google, get an answer, copy-paste it somewhere, and close the tab. That's fine — but it captures maybe 5% of what these tools can do in 2026.
The other 95% looks very different: an assistant that reads your inbox every morning and drafts replies in your voice, an agent that turns a shoebox of receipts into a categorized tax overview, a research analyst that reads 40 sources and hands you a cited report, a pipeline that turns one LinkedIn post into a carousel, a newsletter section, and ad variations automatically, or a system that flags wasted ad spend before your monthly report does.
None of this requires you to become a programmer. It requires three things: a better mental model of what these tools actually are, a small set of skills — prompting, tool selection, connecting AI to your data — and the habit of building systems, not just having conversations.
What you'll be able to do by the end
By the end of this course you'll be able to explain, in plain language, how large language models work and where they fail; write prompts that reliably produce usable output on the first or second try; and choose the right tool for a task instead of defaulting to one chatbot for everything.
You'll also be able to run deep, cited research projects instead of shallow Q&A, use a coding agent for non-coding work like organizing files or processing receipts, connect AI to Gmail, Calendar, Notion, and Drive, build no-code automations with AI steps inside them, use AI properly inside Canva, Meta Ads, and Google Ads, and assemble your own personal "AI operating system" with a 30-day plan.
How to take this course
Do the exercises. Reading about AI is like reading about swimming — every module has "Try it now" blocks designed to take 5–15 minutes, and they only work if you do them in the tool, not in your head.
Use your real work: every exercise works better with your actual emails, files, and campaigns. Generic practice produces generic skills. Expect some friction — tools will ask you to sign in, connect accounts, and grant permissions, but that friction is front-loaded; you do it once and benefit daily.
Budget-wise, you can complete the entire course on free tiers, but the experience is dramatically better with one paid subscription to ChatGPT or Claude — roughly the price of two lunches a month.
How to run it
- Picked a real recurring task. You named a specific weekly (or more frequent) task from your actual work or life, not a hypothetical example.
- Described it in plain language. You wrote down the trigger, steps, and definition of "done" in a few concrete sentences.
- Tested the description with AI. You pasted it into ChatGPT or Claude and confirmed it understood the task well enough to ask a useful follow-up question.
- Saved it somewhere findable. Your task description lives in a folder or note you'll actually return to during later modules.
- Set a realistic budget expectation. You've accepted that the course works on free tiers but is meaningfully better with one paid subscription.
In the field
A freelance bookkeeper picks "sorting monthly client receipts" as her practice task. She describes it as: "Every month, clients email me 20-40 receipts as PDFs and photos; I open each one, note vendor/date/amount, and type it into a spreadsheet by hand." When she runs this through the plain-language test with ChatGPT, it immediately asks whether the receipts follow a consistent format — a question she hadn't thought about — which sharpens her description before she even starts the course.
When not to use it
This exercise assumes you have at least one task that's genuinely repetitive enough to be worth systemizing. If your week is dominated entirely by unique, one-off creative or strategic work with no repeating shape, forcing a "recurring task" framing can feel artificial — in that case, it's fine to pick the closest approximation (like "writing a weekly status update") rather than inventing false repetition, and to treat the rest of the course as building general-purpose skills rather than automating a specific workflow.
Pitfalls and takeaways
- Treating AI purely as a Q&A tool instead of exploring what agents and connected assistants can do
- Skipping the exercises — reading about AI is like reading about swimming
- Practicing on generic, made-up examples instead of your real work
- Underestimating the friction of setup (sign-ins, permissions) — it's front-loaded and pays off daily
- Most people use 5% of what modern AI tools offer; this course covers the other 95%
- You don't need to code — you need mental models, a few skills, and a systems habit
- Hands-on exercises with your real work are the fastest path to competence
Do the work
Goal: Identify one real, recurring task in your week and turn it into the practice material you'll use throughout this entire course, instead of practicing on generic made-up examples. (20 min)
- List your recurring friction points. Write down five tasks you do most weeks that feel tedious, repetitive, or draining — things like sorting receipts, writing similar emails, prepping for recurring meetings, or summarizing reports.
- Pick the one with the clearest shape. Choose the task that has a repeatable structure (same steps most times) rather than the one that's biggest or most dramatic — repeatable tasks are what AI handles best.
- Describe it in plain language. Write 3-4 sentences describing the task exactly as you do it today: what triggers it, what steps you take, and what "done" looks like.
- Ask AI to reflect it back. Paste your description into ChatGPT or Claude and ask: "Based on this description, what's the most time-consuming part of this task, and what would you need from me to help with it?"
- Create your practice folder. Make a dedicated folder (digital or a single note) labeled "AI Course Practice" and save your task description there — this becomes the real material you'll return to in later modules.
Deliverable: A one-page note naming your chosen recurring task, describing it in plain language, and storing it in a findable "AI Course Practice" folder.
Sources
- OpenAI, "GPT-4 Technical Report" (openai.com/research)
- Anthropic, "Introducing Claude" and subsequent model cards (anthropic.com/news)
- McKinsey & Company, "The State of AI" annual survey (mckinsey.com)
- Stanford HAI, "AI Index Report" (aiindex.stanford.edu)