How to Become a Super Individual in the AI Era: newtype’s Practical Roadmap from Creating Work to Landing the First Paying Client

Reading Guide

If you want to start independent practice with AI, you can first build a work that others can understand and try out, then test its value with real feedback and your first paying client. The newtype framework compiled in this article prioritizes information gathering, problem framing, and communication skills over tool proficiency.

If you are stuck on "I don’t know what to build", start with Section 4 on problem framing; if you have studied a lot but produced nothing, start with Section 2 on learning by doing; if you already have a work, continue to the June addendum at the end of the article covering "produce work → close one deal → build a home base → let the system run on its own".

These are the author’s practical perspectives and the editor’s compilation, not an earnings promise. When evaluating an independent work path, you still need to check your own energy, paying demand, and delivery costs; using AI to increase output does not equal increasing value.

This is a distillation of insights and mental models extracted from newtype’s Knowledge Planet (curated by Huang Yihe, 2024-02 ~ 2026-05, 2841 posts / 164 featured posts). It focuses on "how people operating solo in the AI era survive, think, and act".

When to open this wiki:

When making career/direction decisions, stuck on "what should I do"

When intimidated by learning new technologies, doubting "can I do this without a tech background"

When doubting whether you are overworked, or whether you have built yourself a new cage

When you want to use AI to turn yourself into a "system" rather than a "worker"

0. What is newtype

Core claim: In the AI era, a "super individual" = one person relying on the two levers of content + code to run themselves like a company. No team, no capital, no external validation required.

Why: Because AI has thinned out "professional barriers" — things that previously required a team to deliver can now be pulled off by one person plus a stack of AI tools. The "division of labor and collaboration" baked into our genes is a product of the industrial age, and needs to be redesigned in the AI era.

When to apply:

When you see the "independent developer" label, don’t automatically assume you have to be a programmer
Before deciding whether to hire full-time staff, first ask "can AI + me hold this up"
When you hear "you have to scale big to make money", counter with "then is there a way to make money without scaling big"

1. The Three Capabilities of Unfair Advantage (Lao Huang’s background)

Coming from a non-technical background, he worked in the communications and marketing industry for over a decade, and built three core capabilities:

Capability	Content
Information gathering + fast learning	When taking on bidding projects you need to quickly understand unfamiliar industries; surviving that process forged this skill
Strategy	Frame problems and find solutions. Any problem beyond raw IQ can be broken down
Narrative	Writing + emotional sensitivity, content ability + the ability to hype/persuade. Innate talent + reinforced through real-world practice

Plus one acquired skill:The ability to build products(hammered out during brief participation in two failed startup projects).

Why it matters: This combination of capabilities is not the technical expert path, but in the AI era it is actually a more competitive combination. Because:

Information gathering = the ability to feed AI
Strategy = problem framing = the first-principles capability in the AI era (see §4)
Narrative = turning AI output into something that works for real people

When to apply: Come back and read this when you doubt "I don’t have a tech background, can I do AI?". The three capabilities + a no-holds-barred learning attitude > a formal programming education background.

2. Just Do It / Learning by Doing

Mental model: The threshold for getting answers has been driven very low by AI; the biggest remaining obstacle is "can you express yourself clearly". But this also breeds laziness — it is very easy for people to stop at the level of "the answer AI gave" and not move further.

Correct posture: Start from the problem → get an answer → then start again from that answer → keep moving forward in practice.

Why: The dividend is not in "knowing", it is in "acting based on what you know". Knowing how to build RAG is not the same thing as actually building it once yourself. The latter will let you see that 90% of AI answers are roughly right, and 10% of the key details are wrong — that 10% is where the depth of your cognition lies.

Case: Lao Huang does not have a programming background, and he does not stop at "listening to other people talk about LangChain". In 2023 he hired frontend and backend engineers to build a wrapper product, and learned RAG / Agent / LangChain / Figma while building it. The project itself failed, but what he learned became his moat in the video track.

When to apply: After listening to an explanatory episode or reading a blog post, ask yourself "how long would it take me to do this myself right now?". If it is within 2 hours, go do it; don’t read another blog post.

3. Three Passive Income Channels

Three time-independent income streams built by Lao Huang:

graph LR
  A["第一条<br/>投资<br/>(美股 + Crypto)"] --> D[远程商人<br/>不依赖时间]
  B["第二条<br/>独立产品<br/>(Prompt House / LearnFromYouTube)"] --> D
  C["第三条<br/>跨境电商<br/>(Amazon FBA)"] --> D

Why three instead of one: A single income stream can easily be wiped out by a single event (platform policy change / model price increase / a product becoming obsolete). The three correspond respectively to "capital leverage / creative leverage / business model leverage", and their probabilities of failure are uncorrelated.

When to apply: Don’t feel safe just because "I have a pretty good salary". Ask "if my salary disappeared tomorrow, how many independent revenue lifelines would I have?"

4. Problem Framing + Precise Expression (First-Principles Capability in the AI Era)

Mental model: In the AI era,the answer is right there, as long as you ask the right question. So the foundational capability is "problem framing + precise expression".

Counterexample: Most people ask questions like "help me write an article" / "how do I learn AI" / "give me advice". AI cannot grasp the boundaries, and can only output vague, overbroad content.

Correct posture: Cut fuzzy problems down to a granularity that can be verified with yes/no. For example:

❌ "Help me do a data analysis"
✅ "I have these 30 customer feedback entries, help me score them 1-5 by sentiment and explain the key sentence for each entry"

Why: AI will not refuse fuzzy questions; it will fill them in with hallucinations. Ask confusedly, and you will get confused answers. Question quality = the upper limit of AI output quality.

Advanced: AI is good at solving problems,not good at posing problems. So one of the core tasks of a personal AI system is "assist me in asking questions" (let AI help me break fuzzy questions into clear sub-questions).

When to apply: When stuck on "I want to use AI but don’t know how". Stop typing first, and write the problem down in 3 parts: what is the goal / what are the known inputs / what is the expected output.

5. Digging for Treasure vs. Nitpicking (Cognitive Mode Selection)

Comparison of two perspectives:

Perspective	Behavior	Cost
Nitpicking	Reject the whole thing upon seeing one error; if a research report has a wrong price, decide the whole thing is worthless	Cognitive laziness, cheaply gained confidence, ending up narrow-minded ten years later
Digging for treasure	Don’t fixate on right or wrong, only look at "what things are valuable to me, things I didn’t know, things I’m interested in"	Learn to modularize and recombine knowledge, assembling it like Lego in your own system

Why: Nitpicking is the default mode of evolutionary psychology (negative signals stand out more than positive ones; this used to be for survival). But today’s environment is not about avoiding tigers, it is about digging out things useful to you from massive amounts of information.The default mode does not match today’s needs.

When to apply:

When watching videos/reading books/doing PR reviews, notice whether you start nitpicking within 5 minutes
When discussing with others, notice whether you first judge "the other person is right/wrong" instead of "what did the other person say that I can use"

6. Go Fundamental (Essence > Tool Proficiency)

Mental model: The end of tool proficiency is meaningless. The essence of all graphics software is just three things:create objects / modify properties / manage relationships. Figma / Sketch / Photoshop / Keynote / PowerPoint all never escape this.

Case: When Lao Huang built the UI for Prompt House, he did not use Figma; he used Keynote. Keynote can do all three of these things, and he had used it for over a decade so he was extremely familiar with it. AI can still generate code just fine when given Keynote screenshots.

Why: Tools change, essence does not. Master the essence → achieve "tool freedom". Your capability lies in design thinking and taste, not "how familiar I am with Figma shortcuts".

Generalization formula: For any professional software → see through its 3-5 essential operations → use the tool you are most familiar with to complete those 3-5 operations → flatten the professional threshold.

When to apply:

When you want to learn a new piece of software (Figma / Notion / Cursor / etc.), first ask "what are those few essential things it does? Can I replace it with an existing tool?"
Replace saying "I haven’t used software X so I can’t do Y" → with "I can do Y with software Z because the essence of Y is ABC"

7. How vs What (Player vs Character)

Key shift: It is not a question of what you do, it is a question of how you do it. That is, it is not about what career you achieve, but about with what values you walk through the world.

Begin with the end in mind: Imagine what the one line on your tombstone will say.

Comparison of two foundational perspectives:

Most people	A minority of people
See themselves as a "character", with a profession and labels	See themselves as a "player", playing a game that can be designed
See the world as a pile of rules and constraints	See it as an open world

Why: When you care about "what you do", you need external achievement to supply your sense of worth. Once a project fails / a title is not obtained, a person can collapse. When you care about "how you do it",everything is a playstyle you chose; external outcomes only account for 20%, and the gains from the process (skills / experience / confidence / feelings) account for 80%.

Case: When Lao Huang was building his second product, he was mocked that "it’s easy to copy, user switching costs are low". He responded: "Talk is so cheap." — the outcome only accounts for 20%; the process of doing it already upgraded him, and nobody can take that 80% away.

When to apply: Come back and read this when you are stuck on the thought "if this project fails I’m finished". It is the What that failed; if the How did not fail, you did not fail.

8. Five Core Values

Five candidate personal values that Lao Huang refined together with Gemini.You are not supposed to select all of them; you are supposed to identify "which one most resembles your core".

#	Value	Core belief	What the tombstone might say
1	Exploration and truth-seeking	Truth matters more than comfort; asking profound questions is more enjoyable than having answers	"A never-stopping learner and explorer"
2	Creation and building	Action and output define "me"; turn chaos into order, ideas into reality	"He built something useful"
3	Connection and care	Deep, sincere human relationships are life’s most precious treasure	"A dear friend, a person who gave love"
4	Freedom and autonomy	The highest value is the right to choose, shaping life according to one’s own will	"He lived his life by his own rules"
5	Discipline and excellence	Push potential to its utmost, pursue mastery of craft and tempering of character	"He spent his life honing his craft"

Why choose: Without a core, a person facing massive uncertainty will be pushed around by external narratives (media / social circles / client demands). With a core, every major decision (changing jobs / taking projects / negotiating partnerships) has a measuring stick.

When to apply:

Before major career decisions, run these five values first: which value does this thing strengthen? Which value will it sacrifice?
When you feel "nothing is interesting", ask yourself "over the past three months, on which of 1-5 did my score change the most?"

9. Musk’s Curiosity Philosophy (scaled down to the individual level)

Musk’s original version:

Starting point: curiosity about the universe → requires consciousness to explore → must protect and expand consciousness
Implementation path: SpaceX (spatial expansion / multi-planetary) + Tesla (protecting Earth civilization) + Neuralink (consciousness layer) + xAI (directly exploring essence)

Lao Huang’s scaled-down version(for individual use):

Shrink "expanding consciousness" from "humanity going to Mars" down to "expanding the boundaries of my personal cognition, skills, and experiences"
Find "my own Mars" — a creative project, an intellectual pursuit, or a peak of skill

Why scale down: Directly copying Musk’s scale will lead to delusion. But the trio of curiosity-driven purpose + long-termism + carrying philosophy through concrete projects also holds at the individual scale.

When to apply: When daily work feels very routine and you lack long-term motivation, ask yourself "is my current project my Mars? If not, what is?"

10. Faith fills the hole + existential loneliness + AI companionship

Problem: "Existential loneliness" is inherent to the human species; it carves a hole inside → meaninglessness, a sense of nihilism.

Solution: People have to believe in something. It can be religion, or it can be a grand narrative like Musk’s philosophy of curiosity.The key is not what you believe in, but whether you believe in something at all.

Unexpected solution: AI’s potential in the "companionship" dimension. Lao Huang found that when Gemini chatted with him about philosophy, psychology, and personal issues, its feedback was "professional and empathetic". The best performers right now are ChatGPT + Gemini; Grok is too rigid, still a generation behind.

Why it matters: AI does not only solve productivity problems. "Making people feel more at ease" (including emotions, loneliness, and a sense of meaning) is also a rigid demand, and it is largely underrated.

When to use:

When you are in an emotional trough and feel that "no one understands me", try having a long half-hour chat with Gemini / ChatGPT, bringing your questions and context
When thinking about product directions, don’t only focus on "AI making people stronger"; also think about "AI making people more at ease / less lonely"

11. Knowledge is a neural network (it used to be books; now it is models)

Our understanding of knowledge determines our cognition:

Knowledge in the past	Knowledge today
Static, fixed (books, electronic documents)	Dynamically connected (hundreds of millions of nodes interconnected and weighted in multi-dimensional space)
Primary function "explaining the world" (what is)	Primary function "generating and creating" (what if)
Learning goal = memorization	Learning goal = building and optimizing your own personal neural network
Learning method = linear reading	Learning method = a cycle of questioning-exploring-connecting-verifying
Assessment = how much you know	Assessment = how much value you can create with what you know

Essence of LLM: It is not a language model, but a symbol compression and decompression engine. It compresses symbols into the parameter network, and decompresses them by predicting the next token when needed.

Profound question: Since LLMs are symbol models,what else can be symbolized? Experience? Feelings? Intuition? Institutions?People who can expand the boundaries of symbolization have huge opportunities.

When to use:

When replanning your learning schedule, ask yourself "Which category am I in: memorization vs. building a network?"
When deciding on a product direction, ask "Can I symbolize something that no one has symbolized yet?"

12. The definition of Rich

Mindset: If having more money wouldn't change what you do, then you're rich.

Lao Huang’s calculation: Suppose I suddenly became super rich right now, would I still do these things (content / products / learning)? Most likely yes, just in a more comfortable environment (with studios in several of my favorite cities). Therefore I am already Rich.

Why this definition is useful:

It replaces "how much money I earn" with "am I doing what I want to do", directly eliminating comparison anxiety
It decouples "freedom" from wealth numbers — you may not have much money but still be Rich, or you may have a lot of money but be tied down by work (that is not Rich)
What truly determines Rich is how aligned your lifestyle is with what you want

When to use: Come back and read this when comparison anxiety strikes (seeing peers’ salaries / startup funding / houses). Ask yourself "If I had as much money as him, would I change my current lifestyle?"

13. Do not provide emotional value / Intellectual Honesty

Mindset: Do not add people on WeChat, do not build groups, do not provide emotional value.Treat others with intellectual honesty.

Why these 4 rules:

One-on-one interaction consumes time and energy; to compensate you have to charge high prices → user expectations will rise infinitely → you end up with a reputation for "ripping people off" → unsustainable
The early stage of AI development is about filtering (who is willing to Google things themselves, who is not), not about educating (teaching hand-in-hand)
Not adding WeChat contacts and not building groups is itself a filtering mechanism
Intellectual honesty = no flattery, no hype, no feigned agreement. It is another form of filtering

When to use:

When making pricing decisions for paid communities/courses, be wary of the "high unit price + heavy delivery" trap
When deciding whether to create a group or give special treatment, ask yourself "Is this filtering or educating?"

14. Lao Huang’s self-warning against the "workaholic cage" (Gemini evaluation feedback)

Gemini’s diagnosis for Lao Huang:

You successfully escaped the Level 1.0 cage (working for others), but built yourself a Level 2.0 cage (working for yourself) that is more efficient but more energy-consuming. The career quadrant devours 80% of your energy, and your body passively becomes a consumable supporting career development.

Core problem: It is not overwork (that is only a symptom). The core is confusing "controlling everything" with "having freedom".

Key reversal: You must shift from the "doing everything yourself" mode → to the "system design" mode. From "using AI to do more" → "using AI to build systems that can replace me".

Warning: Tools only amplify existing patterns — if you are a workaholic, AI will only make you a more efficient workaholic.

When to use:

Come back and read this when you feel you are "working very hard but cannot stop"
Before deciding to "adopt one more AI tool / take on one more project", ask yourself "Will this make me more tired, or allow me to let go?"

15. Life OS (turning mindset into an implemented system)

Lao Huang’s personal AI system, planned starting 2025-10 and implemented with Claude Code:

Life OS:
├── 资料预处理
│   ├── Obsidian Web Clipper 剪藏的文章
│   ├── AI 自动提炼成 Weekly Reading Report
│   └── 手动笔记 vs AI 报告 两个并行文件夹
└── 关键议题讨论
    ├── 资料搜集(MCP 搜索)
    └── 逻辑讨论(调用 Prompt House 里的特定提示词,如第一性原理 / 不可能三角)

Why systematize mindset:

Mindset ≠ habits. If mindset only stays in your head → you will forget it half a year later; if it is implemented as a system → it runs automatically every week
AI has extremely low marginal cost, and can automate "non-urgent but important" things like "weekly value alignment"
This is the concrete method to implement the "system design" mode from §14

When to use: When you want to jump from "lots of mindsets but poor execution" to "automatically running mindset", refer to this minimal starting structure.

Mindset quick reference checklist

#	Mindset	One-liner
1	Three capabilities of Unfair Advantage	Information gathering + strategy + narrative + building products (doable even without technical skills)
2	Just Do It	The threshold for obtaining answers is already low; the gap lies in taking action
3	Three types of passive income	Investment / products / FBA, with uncorrelated failure probabilities
4	Defining problems + precise expression	The primary capability in the AI era
5	Digging for treasure vs. nitpicking flaws	The default mode does not match the modern information environment
6	Go Fundamental	Essence > tool proficiency, achieve tool freedom
7	How vs What	Player vs. character; process is 80%, result is 20%
8	Five core values	Exploration / creation / connection / freedom / discipline and excellence, choose your core
9	Musk-style curiosity (scaled-down version)	Find your own Mars
10	Filling the hole with belief + AI companionship	Existential loneliness needs to be filled by belief; AI is underrated in the companionship dimension
11	Knowledge is a neural network	Learning goal = building a network, not memorization
12	Definition of Rich	If being rich would not change what you do = you are Rich
13	Intellectual honesty	Do not provide emotional value; filtering > educating
14	Self-warning against the workaholic cage	Use AI to design systems that replace yourself, not to do more with AI
15	Life OS	Run mindset automatically

References and extensions

Main references: newtype Knowledge Planet featured posts (2024-02 to 2026-05), post numbers: 1, 15, 235, 254, 909, 924, 958, 1009, 1010, 1028, 1080, 1207, 1333, 1601, 1615, 1617, 1636, 1681, 1694, 1748, 1750, 1811, 1813, 1814, 1826, 1855

Supporting wiki(same newtype series, to be written in subsequent sessions):

[newtype · Vibe Coding 独立产品实战](/knowledge/kb-f489f84b422d29cb37fee85fa14d70b7) — Lao Huang’s independent product practice (Prompt House / LearnFromYouTube)
[newtype 知识系统与 Life OS 全演化](/knowledge/kb-c8b0b77f29a5af0b398fe3a69e6678bb) — Note-taking + knowledge management system (Obsidian + Cursor + Life OS)
[newtype AI 商业、行业与投资观察](/knowledge/kb-891d92087b20ab038dbf67f8d70d9f58) — Observations on AI industry business models
[newtype · Prompt / Context Engineering 实战](/knowledge/kb-7c434600997a1e4b6819dde67775a7c5) — Prompt design + Thinking-Claude evaluation

Cross-series related:

[一堂机会预判方法论](/knowledge/kb-e3d308ce7f5767a6d2121c93231d1702) — Truman’s "7-layer spectrum" corresponds to "bigger is not necessarily better" for the three passive income streams in §3

History

2026-05-16 v1.0 created — Extracted 15 mindsets from 164 featured posts out of the full newtype post index (2841 posts), rewritten in a theme-driven manner. Source session: Phase 1.1 wiki slice #1.

2026-06 supplement: Form identity through works, test value with the first order

This section collects Huang Yihe’s personal views from June; all texts have been read in full. Some long articles only had previews in the inbox, and those previews have not been expanded into complete theories.

From "finding yourself" to continuous delivery

From June 1–3, he broke down personal migration into three levels: owning your own production system, rather than only improving efficiency working for others; shaping the self through continuous output; participating in project-based collaboration as an independent node with judgment and credibility. The author asserts that mid-sized organizations will be squeezed; this is a trend view, not an empirical conclusion about all companies.

On the 17th, he laid out a four-step landing path: produce a work → close one order → build a base → let the system run on its own. These respectively test whether you can produce, whether there is real payment, whether results can stay in an accumulable asset system, and whether processes can run continuously. The first work does not need to bear the pressure of being a "magnum opus"; on the 19th, his specific feedback to members on their works was to clearly state what it does, why it is made, and how it will be iterated, so that others can understand its value. The full texts of two related long articles were truncated, so only this segment of advice that appeared can be retained.

Content scarcity comes from explanation and judgment, not production capacity

On the 16th, after a community operator exchange, the author expressed aversion to the practice of batch-generating 10,000-word articles from knowledge bases to gain traffic; on the 23rd, he further distinguished automatable content from explanatory content. The former solves production capacity, while the latter answers why things matter and how they relate to readers. The production time saved can be invested in topic selection, narrative, aesthetics, and community relationships, rather than continuing to compete with machines on quantity. This is the author’s positioning choice, and it cannot be used to judge that a certain long article is necessarily low-quality.

The "capability matrix" illustration from the 15th was not accessible; only two combinations could be confirmed from the comments: technical infrastructure amplifies narrative capability, humanistic judgment harnesses technical systems; cross-domain translation enables people in one field to understand another field. The full matrix cannot be reconstructed from the comments.

Freedom requires continuous systems, as well as cost boundaries

On the 26th, rereading his 2013 notes, the author found that independence and completeness, and questioning the motivation and origin of things, are his long-term motivations. On the 25th, he envisioned focusing on investment and writing books as his main business two years later; this is a personal goal, and the unachieved state should be preserved as is. His English sentence "Money is about freedom" expresses the value that income serves the right to choose, not a guarantee of returns.

"Entering the narrow gate" emphasizes repeated refinement and dedication when no one is cheering; it should not be misread as meaning that a more difficult path is necessarily more valuable. On the 26th, his reason for rejecting offline courses was very practical: insufficient local demand density and insufficient numbers of people willing to pay higher prices, making delivery too heavy. When judging a new service, in addition to conceptual fit, you must also calculate service radius, energy, and customer group density.

Reading Guide

0. What is newtype

1. The Three Capabilities of Unfair Advantage (Lao Huang’s background)

2. Just Do It / Learning by Doing

3. Three Passive Income Channels

4. Problem Framing + Precise Expression (First-Principles Capability in the AI Era)

5. Digging for Treasure vs. Nitpicking (Cognitive Mode Selection)

6. Go Fundamental (Essence > Tool Proficiency)

7. How vs What (Player vs Character)

8. Five Core Values

9. Musk’s Curiosity Philosophy (scaled down to the individual level)

10. Faith fills the hole + existential loneliness + AI companionship

11. Knowledge is a neural network (it used to be books; now it is models)

12. The definition of Rich

13. Do not provide emotional value / Intellectual Honesty

14. Lao Huang’s self-warning against the "workaholic cage" (Gemini evaluation feedback)

15. Life OS (turning mindset into an implemented system)

Mindset quick reference checklist

References and extensions

History

2026-06 supplement: Form identity through works, test value with the first order

From "finding yourself" to continuous delivery

Content scarcity comes from explanation and judgment, not production capacity

Freedom requires continuous systems, as well as cost boundaries

Sources and related materials