AI Basics
Build a shared vocabulary and a map of the industry: where AI comes from, how far it's gotten, and what each role in the value chain does.
A Map of AI Terms
Put the high-frequency terms on one map: what each is, which problem it solves, and how it relates to you.
02A Short History of AI
From symbolic rules and machine learning to LLMs. What each paradigm shift replaced.
03Where AI Stands Today
What's stable and usable now, what's still changing fast, and what demos well but isn't reliable in production.
The Four Layers of the AI Industry
Compute, models, platforms, and applications. See what each layer sells, pays for, and builds its moat on.
05How Model Providers Differ
Skip the vendor rankings. Make a real choice by capability, cost, ecosystem, deployment, and compliance.
06Where to Get Models
Official APIs, cloud platforms, model hubs, and aggregators differ in control and responsibility.
LLMs
How large language models work. From the single mechanism of "predict the next token", every strength and every flaw follows.
How LLMs Actually Work
Predicting the next token — that's the whole mechanism. Every strength and every flaw grows out of it.
02LLMs vs. Search Engines
Search hands you the bookshelf; an LLM hands you a conclusion. Answering from memory means there's no source to check.
03Training vs. Inference
One feeds the model, the other uses it. The boundaries of cost and capability sit on either side of this line.
04Where Training Data Comes From
What the corpus is made of decides whose bias the model carries and which parts it's missing.
Six Kinds of Models
Text, image, video, speech synthesis, speech recognition, and embeddings. Input, output, and billing units differ.
06Reasoning vs. Regular Models
Deep thinking means generating a chain of thought before answering. Which tasks justify that extra cost.
07Open vs. Closed Weights
What public weights mean, and when self-hosting actually pays off.
Tokens: The Billing Unit
LLMs don't bill by characters. The price gap between languages comes from how text is tokenized.
09How to Read Params and Benchmarks
Parameter count, context length, and benchmark scores are three different things. Don't mix them.
10Temperature and Randomness
Getting a different answer each time is by design, not a bug. When to turn it off.
What Calling a Model Takes
base_url, model name, and API key. Why everyone is compatible with the OpenAI format.
12Common Call Parameters
max_tokens, top_p, stream, system — what each parameter means for your product.
13Selection and Fallbacks
Tier models by task, and auto-failover when the primary model goes down.
Why Hallucinations Happen
Hallucination is an inevitable byproduct of the generation mechanism, not a bug. You can only defend, not fix.
15Knowledge Cutoff
An LLM doesn't know what happened yesterday unless you feed it yesterday.
16Multimodality: How Images Get Read
Multimodality is a retrofit. How it was added decides how closely the model can look.
Prompting
A prompt isn't an incantation — it's a requirements document. The clearer you write it, the more it gets right.
A Prompt Is Not an Incantation
It's a requirements document. Wherever you didn't specify, the model fills in its own answer.
02The Four Parts of a Prompt
Context, requirements, constraints, and acceptance criteria. Miss one, and it goes off the rails on that one.
03System vs. User Prompts
What to carry in every turn, what to say once, and how priority works.
Why Write Prompts in Markdown
Structured text is parsed more accurately. This isn't a matter of taste.
05Examples Beat Explanations
How few-shot examples work and their traps: how many to give, which ones, and when they hurt.
06Controlling Output Format
From "return JSON" to enforced formats, and the trade-offs when streaming.
A Few High-Leverage Tricks
Asking questions first, reference anchoring, think-before-answering — and what role prompts really do.
08How to Iterate on Prompts
Change one thing at a time and keep a set of test cases. Tuning by feel doesn't work.
09Storing and Versioning Prompts
Prompts scattered through code will eventually become a blob no one dares touch.
Context & RAG
LLMs have no memory. First see how retrieval finds things, then how to fit material into a limited window.
Embeddings and Semantic Similarity
Turn a sentence into coordinates; similar meanings land close together. The whole retrieval stack is built on this.
02Similar Is Not Relevant
The most similar passages aren't necessarily the answer to your question. This is the biggest error source in retrieval.
03Keyword, Vector, and Hybrid Retrieval
Three ways to search, each with blind spots. Real systems almost always mix them.
The Context Window
Everything a single request can hold — that's the size of the table you get.
05Lost in the Middle
The start and end are read clearest; the middle fades as the context grows.
06Conversation Compaction
Long chats quietly drop content. What drops first is usually the constraints set early on.
07How Long-Term Memory Works
Extraction, conflicts, injection. What it costs to make a model recognize you.
RAG: The Three Steps
Retrieve first, then answer. The three stages, and what it actually solves.
09Chunking: RAG's First Do-or-Die Step
Chunk too small and you lose context; too big and the key point gets diluted. This step decides everything.
10Where RAG Goes Wrong
Chunking, recall, and reranking — every stage can fail, each with its own symptoms.
11Fine-tuning or RAG
To add knowledge, it's almost always RAG. When fine-tuning is actually the right call.
Agents & Skills
From chat to getting things done — the layers in between: tool calling, loops, Skills, and MCP.
Model, Agent, Application
A diagram that separates the three layers. Confuse them and you'll treat a product problem as a model problem.
02Tool Calling
The model can't do anything by itself. You lend it your hands; only then can it touch the world.
03The Agent Loop
Think, act, observe, think again. That's how anything running dozens of rounds spins.
Skills and MCP
One carries knowledge, the other carries interfaces. They solve different problems; don't mix them up.
05Multi-Agent Collaboration
When it's worth splitting into several, and where the added cost shows up.
06Layers of Prompts
System prompts, standing rules, and one-off instructions each govern a slice, with different priorities.
Three Ways Agents Drift
Goal drift, tool misuse, and loops that don't converge — each with its own way to stop it.
08Permissions and Human Checkpoints
Which steps to gate so it can't go and delete the database.
09Evaluating Agents
No eval means running blind. A minimal acceptance plan you can actually run.
Cost & Safety
How the token bill is calculated, how to defend against prompt injection, and which judgments should never be handed over.
How a Call Is Billed
Input, output, and cache are priced differently — by several times. Read the price sheet before choosing a model.
02Why Image Generation Costs 10x
How image tokens are counted, and how a resolution change jumps the bill up a tier.
03Cache Hits and Saving Money
A cache hits only when the prefix is unchanged. Reorder your prompt and the bill can halve.
Prompt Injection
The model can't tell which words are instructions and which are data. This is a flaw at the principle level.
05Where Conversation Data Goes
Web, API, and enterprise tiers route your data to different places.
06Common Flaws in AI Code
Hardcoded secrets, broken access control, injection — the ones AI leaves most often.
做产品
Basics
What everyone in this trade is assumed to know. Skip if you already do.
What a PM Does
The boundaries of the role, and where it sits in a company's org chart.
02The Product Workflow
The stages from idea to launch, and when each role shows up.
03Product Types & Directions
Consumer, enterprise, platform, tool, content — each judges success differently.
04The Internet Industry Map
Where e-commerce, social, SaaS, content, and tools stand today.
05Common Business Models
Where the money comes from: ads, subscriptions, commissions, freemium — and how to validate each.
Understand Tech
Frontend, backend, APIs, databases, auth, deploy, security — a minimal technical baseline.
07Glossary
The jargon of both product and tech — look it up so you're not lost in meetings.
Find Problems
Where needs come from, which are real, which are worth doing.
What Is a Need
What users say, what they actually want, and what you plan to give them — three different things.
02Four Sources of Needs
User feedback, business goals, data anomalies, competitor moves — each with different credibility.
03Real vs. Fake Needs
Telling apart what they say they want from what they'll actually use.
Ask About Behavior, Not Intention
Ask how they did it last time, not whether they'd use it.
05Find the Existing Workaround
How are they coping today? No workaround means the problem isn't painful enough.
06Is It Worth Doing
Frequency, pain, audience size — if all three are low, don't.
Scene Anchor
Replace a vague "users find it inconvenient" with a specific person in a specific moment.
08How to Run a User Interview
Open questions, probe for details, avoid leading — a reusable interview guide.
09Five Whys
Trace a surface request to its root cause before deciding whether to build it.
10Backlog & Prioritization
KANO categories plus a sorting rule — not whoever shouts loudest.
Define Product
Who it's for, what it solves, where the boundary lies.
A Definition Has Three Parts
Target users, core value, feature boundaries — miss any one and you'll waver later.
02The Product Lifecycle
Exploration, growth, maturity, decline — each stage has different priorities.
03How the Business Model Connects
Your product definition has to answer where the money comes from, or it stalls after launch.
One-Sentence Product
If you can't say in one sentence what you make and for whom, don't let AI start yet.
05Single Thread
Find the product's main task; every other feature is a supporting role.
06Narrow the Audience
Built for everyone means built for no one — serve a small group first.
07The Not-Doing List
Writing down what this version won't do is more powerful than writing what it will.
Design Structure
Information architecture, page maps, data models. How the product is organized.
What Is Information Architecture
Let users always know where they are, where they can go, and how to get back.
02Four Organizational Structures
Hierarchy, matrix, linear, organic — different content suits different structures.
03Web vs. Mobile Differences
Screen size, nav position, depth tolerance all differ — structure can't be copy-pasted.
04Four Navigation Forms
Left-only, top-only, left+top, top+left — depends on how many top-level modules you have.
One Screen, One Job
Each page serves one main task; everything else steps aside.
06Grouping & Naming
How you group things is how users understand the product.
07Wide and Shallow
Five to nine top-level items, main paths no deeper than three levels — go deeper and nobody arrives.
Content Inventory
List every piece of information to show before deciding how many pages to split.
09Card Sorting
Write features on cards and let users group them — more accurate than guessing alone.
10Page Map
Draw every page and the jumps between them before you start building.
11Three Structure Diagrams
Feature structure, information structure, product structure — draw each separately, don't mash them up.
12Data Model First
Get entities and relationships wrong, and no UI tweak will fix it.
Design Interaction
Paths, states, feedback, edge cases, permissions. Every moment that goes wrong.
Interaction Handles Behavior
IA decides where things go; interaction decides how users move across them.
02Common Components
Dropdowns, steppers, pagination, drawers — know them all before choosing.
03Five Types of Hints
Badge, toast, banner, announcement, push — from weakest to strongest.
Four States Complete
Empty, loading, error, normal. AI writes only the last one by default.
05Every Action Needs Feedback
After clicking, how does the user know it took effect?
06Undoable Beats Confirm
Give dangerous actions one chance to undo, instead of three confirmation dialogs.
07Permission as View
Features you can't access shouldn't appear and then error.
Map the User Journey
Break a complete task into ordered steps, mark each step's exit.
09State Checklist
Ask each component what states it can appear in, before you start writing.
10First-Run Design
On first open, with zero data, what's on the screen?
11Form Restraint
Every extra field costs you a batch of people willing to finish.
12How to Write Hints
State the situation plus the next step. No exclamation marks, no trailing punctuation.
Design Interface
Layout, hierarchy, color, typography, design systems.
Where Visual Hierarchy Comes From
Size, position, whitespace, contrast — four things decide where the eye lands.
02Color Basics
Hue, lightness, saturation — and why neutral colors fill 90% of an interface.
03Typography Basics
Font size, line height, weight, tracking — Chinese and English use different parameters.
One Spacing Scale
Lock a set of fixed values; don't eyeball each spot.
05Type Scale
Three to four sizes is enough — more and the hierarchy scatters.
06One Accent Color
Keep one color; leave the rest to neutral grays.
07Whitespace First
Try deleting dividers and borders first; separate with spacing instead.
08Alignment Axis
Align every element on the page to a few axes — it instantly looks clean.
Components First, Then Pages
Without unified components, the page will always look stitched together.
10Build a Minimal Design System
Color, spacing, type, radius — four sets of variables govern the whole project.
11How to Do Responsive
Don't dump desktop components onto mobile — think through the trade-offs at each breakpoint first.
Work with AI
Translating your judgment into something AI can execute.
What It Can and Can't Do
It can fill in implementation, not judgment. Know the difference to know what's yours.
02What Context Is
It only knows what you give it; memory doesn't carry across sessions automatically.
03Spec, Prompt, Rules — Divided
One-offs go in specs, recurring ones in rules, temporary ones in prompts.
Spec Before Code
Have it write what to build first, then have it write code.
05One Thing at a Time
Break it into small tasks; don't ask for a whole app at once.
06Reference Anchor
Give it a concrete reference; ten times more effective than adjectives.
Three-Part Prompt
Goal, constraints, acceptance criteria — miss any one and it drifts.
08Context Budget
What to include and what not to — more important than how much.
09Prototype to Code
Static page first to confirm the look, then wire real data, then cover the edge cases.
10Distill the Rules
Rules you re-state constantly go into CLAUDE.md and Skills.
Validate & Iterate
Check before launch, read data after launch.
Usable, Useful, Used
Three layers of validation, each checking something different.
02How to Build a Metrics System
Break down from the North Star to the layer a single feature can move.
Track Before Launch
A feature shipped without tracking is a feature that didn't ship.
04One North Star
Pick one metric that means the product is getting better; the rest are reference.
05Judge by Real Data
Small samples and your own team's feelings don't count.
The Five-Second Test
Can a stranger look for five seconds and say what this is?
07Launch Checklist
Function, experience, data, security — check item by item before launch.
08Stress-Test with Real Data
Run it once with absurdly long content and once with zero rows.
09Security Check for AI Code
Hardcoded secrets, broken access control, injection — the traps AI leaves most often.
10Splitting the North Star
Split one top-line metric into several paths you can each act on.