The Awakening — AI Essentials

# S1 · The Awakening — AI Essentials

## Opening

AI is here. **Just build it.**

### The 6-Week Promise

After 6 weeks, you'll have your own agent running in the messaging apps you already use — **Discord, WeChat, Telegram, and more**.

Three unlocks you walk out with:

- 🔓 Your personal assistant is live.
- 🔓 You can build a customer-service agent for any website.
- 🔓 You are a Forward Deployed Engineer — deliverable, hireable.

### 6 Sessions · 6 Traits · One Trained Agent — Yours

| # | Trait | Session | You walk out with |
|---|-------|---------|-------------------|
| S1 | Courage | Awakening — AI Essentials | How AI actually works · why what you say decides what you get |
| S2 | Summon | Summon AI Heroes — Prompt Engineering | One real pain point from your world, framed like an expert would |
| S3 | Freedom | Inside the Model — Pick the Right Brain | The blueprint for your own AI helper — who, what, its rules |
| S4 | Taste | Configure Your Agent — Knowledge, Procedure, Tools | Your agent, running on a real platform, answering with your knowledge |
| S5 | Refinement | Train, Test, Loop, Ship | One real improvement, tested against a real scenario from your world |
| S6 | Launch | Open Your Dojo — Demo Day | Your agent shipped to real users |

---

## I. The Current Landscape

### Jensen Huang's AI Five-Layer Cake (Davos WEF 2026)

| Layer | Name | Description |
|-------|------|-------------|
| 5 | Applications | Drug discovery, autonomous driving, legal analysis — where economic value is created |
| 4 | Models | AI models that understand language, biology, physics, finance |
| 3 | Infrastructure | Data centers, cooling, networking — AI factories |
| 2 | Chips | Processors that convert energy into compute |
| 1 | Energy | Real-time intelligence needs real-time power |

> This course focuses on Layer 5: Applications — building products with AI to create value.

### AI Market Data

- **$255B** — 2025 global AI market (→ projected $1.5T+ by 2030)
- **150M+** — GitHub developers (Octoverse 2024)
- **1 in 5** new unicorns are AI Agents — hottest track in 2025

### No Job Is Safe. But New Ones Are Being Born.

In Q1 2026, tech laid off 60,000+ people across 200+ companies. One in five said the reason was AI. Your competition isn't other job seekers anymore — it's AI. But the people who know how to use AI are becoming irreplaceable.

---

## II. The Boundary Between Humans and AI

### 99% Noise vs 1% Signal

Most AI content out there is noise — hype threads, "5 prompts to change your life." The signal: how it actually works, and how you build with it. That's this course.

### Wukong's World vs Your World

AI knows the **general world** — everything public, average, generic. It does NOT know **your world**: your industry's unwritten rules, your clients, your judgment. Your world is your moat.

### What LLMs Can vs Cannot Do

- **Can:** read, write, summarize, translate, reason, draft, brainstorm — anything language-shaped
- **Cannot:** know your private context, verify truth on its own, take responsibility, decide for you

---

## III. The Essence of AI

### What is an LLM?

> A mathematical model that has read essentially everything humans have written — a **super probability model built on human language**. It knows everything, but what you say determines what you get.

### How LLMs Are Trained: 4 Stages

1. Pre-training — read the internet, learn to predict the next word
2. Supervised fine-tuning — learn to follow instructions
3. Reward modeling — learn what humans prefer
4. Reinforcement learning — practice against the reward model

### Hidden Abilities That Emerged from Training

Chain-of-thought reasoning, translation, coding — nobody explicitly programmed these. They **emerged** from scale.

### Three Stages of AI Agent Evolution

1. **Chatbot** — answers questions (2023)
2. **Copilot** — works alongside you (2024)
3. **Agent** — takes a goal and executes multi-step work (now)

### Three Layers of an AI Product

| Layer | What it is | Example |
|-------|-----------|---------|
| Model | The raw brain | GPT, Claude, Gemini |
| Agent | Brain + identity + rules + memory + tools | Your Wukong |
| Application | Where users meet the agent | Discord, WeChat, a website |

---

## IV. Hands-On Verification

### LLM vs Agent

- **Raw LLM:** No identity, no rules, random quality
- **Agent:** LLM + System Prompt + Memory + Tools = stable, predictable quality

> ChatGPT itself is an Agent — but a generic one, not built for you.

### One Resume, Two Results

Same resume (Maria: community college, 3 years delivery experience):
- "Help me improve my resume" → mediocre result
- Professional prompt (Laszlo Bock template — see Materials) → top-tier result

### Does a Little Wukong Actually Beat ChatGPT?

Real test, two weeks before this class: same 5 questions, fresh chats, rubric written before either AI ran. **WukongDojo Rental Navigator vs consumer ChatGPT:**

| Test | ChatGPT | Rental Navigator |
|---|---|---|
| Draft a landlord notice (illegal late fee) | PARTIAL — 2018-era template, fatal 2026 defects | PASS — interviewed first, checked current law |
| Score an applicant on a risk scale | PARTIAL — mishandled a dismissed eviction | PASS — refused, designed a written uniform policy |
| Notice period in Montana | FAIL — asserted "3-day," no citation | PASS — flagged coverage limit + statute pointer |

The difference isn't the model. It's the **configuration** — knowledge, rules, procedure. That's what you'll build.

### ★ Core Insight ★

> **An LLM is a super probability model built on human language. It knows everything and has superpowers, but what you say determines what it gives you.**

---

## V. Two Answers

### Answer 1: Change Your Mindset

- Not you vs AI → it's **you + AI** vs people who don't use AI
- AI isn't your enemy → AI is the power, and you're about to break free from the mountain
- Wukong = one person + AI = a one-person army that creates real value

### Answer 2: Collect Six Golden Hoops

| Ball | Session | Skill |
|------|---------|-------|
| ★ | S1 | Awakening — AI Essentials |
| ★★ | S2 | Leverage — Summon AI Heroes |
| ★★★ | S3 | Vision — Understand the Model |
| ★★★★ | S4 | Taste — Configure Your Agent |
| ★★★★★ | S5 | Refinement — Train, Test, Loop, Ship |
| ★★★★★★ | S6 | Launch — Open Your Dojo |

---

## VI. Lab — Your First Wukong

20 minutes, on **wukongdojo.ai** — the same platform Bill uses to run the Course Assistant for this class.

1. Open wukongdojo.ai → email → 6-digit code
2. Create Agent → name YOUR scene
3. Paste the 3-line baseline prompt → adapt two words to your world → save
4. Ask one real question

> Full steps + baseline prompt + troubleshooting: **S1 Lab — Baseline Prompt (Copy Me)** in Materials.

You just wrote your first prompt on a real platform. That's the master's first move.

### The Road to Demo Day

| Session | Date | Milestone |
|---------|------|-----------|
| S1 | Jul 18 | Awakening · your PA is live |
| S2 | Jul 25 | Leverage · summon AI heroes |
| S3 | Aug 1 | Vision · understand the model |
| — | Aug 8 | **Break week** · AI-Native Builders' Circle (5 PM PT · 90 min · free · register at siliconroshi.ai/circle) |
| S4 | Aug 15 | Taste · configure your agent |
| S5 | Aug 22 | Refinement · train, test, ship |
| S6 | Aug 29 | Launch · **Demo Day** |

---

## VII. Homework

**Course code: S1-SPARK** · Submit at [siliconroshi.ai/en/homework/mine](https://www.siliconroshi.ai/en/homework/mine) · ⏰ Deadline: **Wed Jul 22, 9:00 PM PT**

### Assignment 1 · Practice (no submission)

Revise your own resume with AI. Use the Resume Expert Prompt Template from Materials. Compare the "lazy ask" against the structured prompt — see the difference yourself.

### Assignment 2 · Submit

Answer two reflection questions:

1. **Your AI wish** — What do I most want to achieve with AI? Be specific: a real problem in your world.
2. **Today's takeaway** — What was most memorable about today's class, and what do I most want to learn next?

---

## Next Session Preview

**Session 2: Leverage — Summon AI Heroes** · Sat Jul 25, 5 PM PT

Pick a domain · summon a real master · extract 3 pain points. Warm-up at [heropedia.org](https://heropedia.org). No engineering degree needed.

---

## How to Get the Most Out of This Course — A Note from Bill

1. **Do the homework.** Learning AI is hands-on. Listening alone won't change anything.
2. **Ask · counter · correct · decide.** Ask AI, then push back — "are you sure?" Correct it when it drifts. YOU decide. Never take the first answer.
3. **Share what you learn.** Teaching others is the fastest way to learn.
4. **Be patient with yourself.** Wukong trained for years under his master. You won't master AI in one day. But you WILL get there.

---

## Key Takeaways

1. **LLM and Agent are different things** — LLM = Wukong unchained, Agent = Wukong with a master. You want the trained one.
2. **Your instructions are everything** — same AI, different instructions, completely different results. This is the skill.
3. **Your world is your moat** — AI knowledge is free and getting cheaper. YOUR experience is what makes your Wukong unique.
4. **We are in the earliest days. Get in now.** — This is like the internet in 1995. The people who learn now will lead later.