Chen ZJ
"Stay hungry, stay foolish" — I like to translate it as: Seek wisdom eagerly, enjoy being foolish.
Interests: Tennis enthusiast who admires Federer, boxer capable of self-defense, level-10 scholar in Werewolf reasoning.
From Information Noise to Deep Thinking
The daily volume of information is enough to write a novel, but most people’s conclusion after reading it all is: “Oh, I see” — and then nothing changes.
The core of deep thinking is not thinking more, but thinking more accurately. Being able to write in the noise, being able to judge direction in trends — not by inspiration, but by muscle. Trainable, iterable, reviewable.
“ZJ Deeper” is where we train this muscle. Since 2017, through entrepreneurship, I’ve met a group of highly cognitive people who all share one trait: they don’t just read a lot, they think with structure.
Information is raw material, insight is the product. The goal of this circle is to refine the former into the latter together — not to add another community label, but to have a place where we can persist in “thinking deeper.”
Who Am I
Chen ZJ. My name sounds similar to ByteDance in Chinese, but I used this pen name before they changed their name — as for why I didn’t register the trademark first, this is the only decision in my career I don’t want to deeply review.
I’ve always adhered to deep thinking. The companies I founded — DeepQuant (quantitative investment), DeepAI (AI products) — weren’t deliberately named; they naturally grew from my working style: no superficial work, dig deep.
From “Bigger and Bigger” to “More and More Refined”
Over the past decade-plus, I’ve worked in the tech industry doing: investment, entrepreneurship, research, consulting, quantitative trading, training, product, growth, operations — basically the “full-stack tech worker” version. I’ve changed roles many times and stepped in more pits than I’ve drawn pies.
But one thread has never broken: Less relying on epiphany, more on systems. Breaking down every win and loss into reviewable structures, so the next round can be bet more clearly.
In 2023, I put all my energy into AI products, with no energy to show off outside. In 2026, I picked up the pen again, personally stepping up as a sniper. The goal shifted from “bigger and bigger” to “more and more refined.”
Deep Thinking Methodology
From AI, I learned one counterintuitive thing: Good thinking doesn’t rely on inspiration, it relies on process.
High-quality input → Multi-angle processing → Actionable outputThree words: Select, Think, Write.
- Select: Information filtering is the starting point of the moat. Feed garbage in, and what comes out is definitely a processed version of garbage.
- Think: It’s not “whether you thought,” but “what framework you used to think.” The same thing, with a different mental model, can lead to completely opposite conclusions.
- Write: Output forces input. “Thought it through” that can’t be written out is mostly fake clarity.
This process isn’t sexy, but it can run long-term.
What I’m Really Doing
This is where I’m investing the most, and where I most want to find fellow travelers.
Org Hacking
Growth hackers rewrote the logic of marketing with data and experiments. Org hacking aims to rewrite the logic of “how a company operates” with AI and engineering thinking.
Traditional management asks “how to manage a group of people well.” Org hacking asks: “Does this thing really need a group of people?”
- Hand routine judgments to AI, leave creative decisions to people
- Replace middle layers with Agents, not use middle layers to manage Agents
- A well-designed workflow can let three people do the work of ten — and with better quality
DeepAI is where I’m using my own company as a lab to validate this logic.
AI-Native Thinking
Using AI and AI-native are two different things.
Using AI is: “Let me ask ChatGPT about this problem.” AI-native is: “My entire workflow assumes AI is present from the start.”
AI-native people don’t ask “what can AI help me do,” but ask: “Which parts must I do myself, and which parts are better handed to AI?” The answer to this question needs recalibration every six months.
AI Enhancement Path
It’s not “learning to use AI” — that’s the starting point, not the endpoint. AI enhancement path asks: How can your judgment, professional depth, and network form a multiplicative relationship with AI capabilities, not additive?
An AI-native person + deep professional knowledge = moat. A person who only knows how to use AI tools = replaceable.
Systems Design Thinking
Whether it’s products, organizations, or personal workflows, the core question of systems design is always:
Can this system run on its own when I’m not watching?
Good systems have feedback loops, self-correction mechanisms, clear interfaces — not dependent on heroes, not on luck, not on “everyone working hard.” DeepQuant’s quantitative strategies and DeepAI’s product architecture are both things I’m doing with the same systems design thinking.
Questions I’m Really Thinking About
I don’t make a “areas of interest” list, only list things I really can’t sleep thinking about:
In the AI wave, who’s cheering and who’s eating meat? Trends and opportunities aren’t the same thing. Knowing AI is changing the world ≠ knowing how you should position yourself; there’s a “judgment” in between.
What kind of products can survive cycles? “Viral hits” are easy, “durable” is hard. I’m looking for that kind of logic that’s a bit slow but goes far.
What’s the minimum viable unit of an organization? After AI Agents reorganize, how many people does a “company” minimally need? Where are the boundaries? I’m experimenting with my own company.
What will the most valuable people look like in the future? Not “people who can use AI” — that threshold is too low. What makes a person still irreplaceable in the AI era?
Seeking Co-Builders
If you’re doing any of the following, we likely have things to talk about:
- AI products or tools: Especially those facing professional scenarios with deep data or vertical barriers
- Organizational restructuring or team AI-ification: Exploring how to do bigger things with smaller teams
- Quantitative investment / information finance: Data-driven judgment systems, DeepQuant’s core direction
- AI-native workflows: Not “using AI to accelerate existing processes,” but “redesigning processes”
- Content × Smart Media: Transforming deep thinking into automatically propagating matrices
I’m not looking for “resource exchange” type collaborations, only people who are direction-aligned, capability-complementary, and willing to talk deeply.
Who Is This Circle For?
It’s for you if:
- You feel “I’ve read a lot, but haven’t thought through much,” and want to change that
- You’re in AI / tech / investment / entrepreneurship, but don’t want to just run with the crowd
- You like dense exchanges, not the mutual cheerleading kind
- You’re willing to seriously output, not just consume others’ views
- You still have curiosity about the world and some standards for yourself
Might not be for you if:
- You only want conclusions, don’t care about the reasoning process
- You treat “likes” as a substitute for thinking
What You’ll See Here
- Weekly reviews: Real observations, judgments, sometimes confusion — not packaged as correct answers
- Methodology breakdowns: Explaining useful frameworks clearly, not copying, only refining
- Industry insights: AI, Web3, information quantification, product growth — not chasing hot topics, chasing structure
- Org hacking experiment logs: Experiments I’m doing with my own company, both successes and failures will be written
- Error logs: Judgment mistakes are more valuable than success stories, I don’t hide them
Cognition can be stacked — provided you find people who can spar with each other.
If you’ve read this far and haven’t closed the page, we can probably talk.
ZJ Deeper, Since 2017. Welcome in, let’s make “thinking deeper” a habit together.
