# AI Levels 2026: Cheat Sheet

Based on my video "Stop Using AI Like a Beginner (2026 Framework)".

---

## Level 1: AI awareness (chat tools and AI wrappers)

### ChatGPT
- **Deep Research:** use it when an answer needs more than the first page of Google.
  - Product comparison: name two products and describe your own use case in detail.
  - Sales research: look up a prospect, a company or its decision-makers before a meeting. Start by running it on yourself.
- **Privacy setting:** Settings > Data controls > turn off "Improve the model for everyone".
- **Custom GPTs:**
  - FAQ bot: give it a short text document about your services and policies, test it, publish it and share the link. The people who use it don't need to pay.
  - Internal use: store your team's tone and format rules so everyone can check their own work.
  - Questions to test it with:
    - What are the services that this company offers?
    - What are the return policies?
    - Do you offer student discounts?

### Gemini (large memory, NanoBanana Pro images)
- Report prompt:
  > Here is a report from Meta. I want you to go into this report, analyze it, and get all of this information out there.
- Check any number that looks wrong against the page and line Gemini cites.
- Image prompt (if Gemini doesn't use NanoBanana Pro, click Tools > Create images):
  > Can you create an image of the financial KPIs year on year that you extracted from that report, please?
- Slide prompt:
  > Now I want you to create a slide using the KPI data.
- Infographics: ask for a version for kids, then a version for an expert. The only change is the wording you ask for.

### Claude (artifacts)
- Go to apps and websites, then describe the app:
  > I just want you to generate a very simple coin toss game with heads and tails.
- Then ask for more features (multiplayer, scores). The same approach works for dashboards and client portals.

### AI wrappers
- Apps built on someone else's model. Perplexity adds web search to a model, and Lovable builds apps from a description.

---

## Level 2: AI integration (workflows)

1. **Write the process down first (SOPs).** You can't automate steps nobody has written down.
2. **Automate without AI where possible.** Example, invoice reminders in Make.com:
   - Every invoice you send goes into a Google Sheet.
   - Until you remove the row, the client gets an email after 5, 7, 14, 21, 30 and 60 days.
   - Each email is firmer than the last.
3. **AI-enhanced workflow:** add one AI step where a person would normally decide yes or no. Example: an n8n workflow collects job posts with AI keywords, a ChatGPT step judges each one, and the good ones go into a database.
4. **Agentic workflow:** the AI chooses which tools to use and what to do. Example: a WhatsApp agent for a coach asks leads about their motivation, budget and start date, then sends the coach a summary. Keep the task narrow, because agents don't always do the same thing every time.

---

## Level 3: AI adoption (working with agents)

- Antigravity can use your browser, the web and the files on your computer.
- Browsing prompt:
  > Can you go to the Apple website and give me the latest products that they released, please? Just create a list for me after you browse it.
- Other uses from the video: test your website for accessibility, responsiveness and bugs, build dashboards, and find and rank the people who engage most with your LinkedIn posts.
- You don't need to write code. You need to understand the concepts (processing, batching, APIs), check that the output is right and follow good practice.

---

## The five core AI skills (as described by Nate B. Jones)

| Skill | The question to ask |
|---|---|
| 1. AI strategy | Should I use AI in this workflow at all? |
| 2. Prompting | Which part of this prompt is vague and could cause problems? How do I test that? |
| 3. Workflow integration | If yes: which tools, how do they connect, where are the limits, and where does a human decide? |
| 4. Human evaluation and taste | Who reviews the output so it doesn't become AI slop? |
| 5. Ethics | Where are the guardrails, and why should customers and partners trust what I build? |