First, the answer: How did Zhang Zala turn everyday learning into Xiaohongshu content?
The borrowable part of this historical case is reducing the production burden of individual pieces of content: start from something you just learned, just solved, or just made, explain the process clearly, and then adjust positioning based on real feedback. It is suitable for people who have full-time jobs and want to continuously share AI learning and small-product practice; personal experience is not a follower-growth formula, nor can it directly replace the editorial review process for a team’s commercial content.
Source re-verification: 2026-10-25. Figures in the original draft such as 260,000 and 50,000 were self-reported at the time; there is also a supplementary note at the end dated 2026-05-31 about 300,000 followers, which is not today’s follower count. That self-report explicitly said AI was not used for content creation, while the early tool list included AI-assisted tools; please preserve the timing and wording of both materials separately, and do not piece them together into the conclusion that “she has always used a certain AI pipeline.” Statements such as “no more than 30 minutes per post” are the author’s method constraints, not a productivity guarantee for everyone.
Jump to the main text based on your current question
- If you always feel you have no time, start with How to lower the production threshold for individual content pieces.
- If you don’t know what to write about, start with Use continuous output to refine positioning.
- If you have learned a lot but produced no output, read Merge learning in public with output.
- If you plan to operate multiple platforms at the same time, read The historical division of labor among Xiaohongshu, X, and LinkedIn, then adjust according to your own audience.
A small exercise you can do first
This site’s suggestion: choose one problem you actually solved this week, and use a short post to explain “what you encountered, what you tried, what the result and limitations were,” preserving necessary sources or before-and-after comparisons. Before posting, check whether it leaks other people’s information and whether unverified claims about tools are written as personal tests. Afterward, record production time, specific questions raised by readers, and whether it helped you continue improving the work; do not judge whether the method works based on a single like or follower change. This is an exercise design, with no newly added follower-growth or monetization experiment.
If you need to find input materials, continue reading Installing and first reading of follow-builders; if you already have a small tool, continue reading Zhang Zala’s Vibe Coding practice. The text below preserves the evidentiary boundaries of the original case, supplementary notes, and comments.
Numerical confidence: Xiaohongshu 260,000+ followers / Twitter 50,000+ followers(gained in 2025) / LinkedIn 14,000 / 150 internal company blog posts read by 24,000 colleagues Sustained cadence: 2022-04 to present About 600 days, 300+ pieces of content posted, average1 post every 2 days, sustained for 2+ years Core counter-consensus:Amateur interest + no ads + no lighting and no makeup + single post ≤ 30 minutes + no trend research + no scriptwriting — a falsifiable SOP for building a 260,000-follower scale through hard output in spare time
When to open
| You are thinking | Turn to |
|---|---|
| “Work is so busy, where would I find time to make content?” | §1 Keep the output threshold low |
| “How do I find positioning / what should I post?” | §2 Positioning is forged through experimentation |
| “How much money can making content earn?” | §3 Money is a minor goal / 6 more valuable returns |
| “How to expand my network (especially for introverts)” | §4 Inbound > Outbound |
| “I’ve been doing it for months and nobody reads it; should I give up?” | §5 Post more + keep a light mindset |
| “How to make followers grow faster” | §6 The content market is not a racetrack; it is made through action |
| “What equipment / tools should I use to record videos?” | §7 Tool stack |
1. Lower the output threshold (turn content creation into entertainment)
1.1 Core mindset
“My mindset toward making content is not ‘work is already so tiring, I don’t want to make content,’ but rather ‘I’m off work, let me post something to relax.’”
Counter-consensus: Many people treat content creation as a second job(heavy investment / scriptwriting / trend research / elaborate cover refinement) → when work is tiring, they cannot keep it up. Zara does the opposite:Treat content as entertainment(equivalent to scrolling short videos / playing games) → no matter how tiring work is, it does not affect posting.
1.2 Seven downgrades for a single post ≤ 30 minutes
| Dimension | Typical blogger approach | Zara’s approach |
|---|---|---|
| Time budget | 1–3 hours | ≤ 30 minutes(post when convenient) |
| Scriptwriting | Write a script | Never writes a script; speaks off the cuff |
| Recording environment | Makeup + lighting + recording studio | No makeup, no lighting, recorded on phone |
| Video editing | PR / FCPX fine editing | CapCut AI subtitle editing, done in 10 minutes |
| Soundtrack / transitions / sound effects | Add all of them | Add none |
| Cover | Design a polished image | Casual screenshot, no retouching |
| Topic selection | Research trends + ride viral hits | Record everyday study, life, and ideas in notes; topics come from notes |
1.3 Use fragmented time
| Time slot | What can be done |
|---|---|
| Waiting for the elevator | Write 1 image-text post (5 minutes) |
| Commute | Record a video draft |
| Queuing in the cafeteria | Organize topic notes |
| On the way home from work | Edit (CapCut opens ready to use) |
Key insight: “The only sustainable output method for working professionals = lower the output threshold, keep it lightweight and quick,post casually. Once it becomes heavy, you won’t be able to keep going.”
2. Positioning is forged through experimentation, not thought up
2.1 Counter-consensus quote
“Positioning is told to you by your audience, not thought up by yourself.” — Crossroads podcast with Koji Yang Yuancheng
Many bloggers agonize for months over “what positioning should I choose” without ever writing. Zara’s approach:Post first, see what catches fire, iterate.
2.2 The evolution of Zara’s own positioning
- Early stage: miscellaneous product PM talk alongside full-time work
- Middle stage: discovered topics related to “liberal arts student switching to AI” tended to catch fire → increased this angle
- Later stage (2025–2026):AI product manager / Vibe coder / independent development in practice
Every positioning adjustment was driven by user feedback, not planned in advance.
2.3 SOP for how to “experiment your way to” positioning
1. Post first (topics according to mood, no fixed theme)
2. Observe which posts are recommended organically / get many comments / drive fast follower growth
3. Increase frequency along “topics that tend to catch fire,” and don’t force yourself to post about things you’re not good at
4. Review every six months and state positioning in one sentence (Zara: “AI product manager + liberal arts vibe coder”)
5. Positioning is not thought up; it is posted into existence
3. Six returns from content more valuable than money
“If you think making content is only for money, then you’re thinking too small.”
Zara has made content for 4 years and has never taken a single ad / earned a single cent. But it brought:
| # | Return | Real example |
|---|---|---|
| 1 | Job opportunities | Every job came from bosses who had seen her content — joining VC (intern reporter article), ByteDance (work connecting U.S. and Chinese entrepreneurship), an AI team (early AI experimentation) |
| 2 | Connections | Entrepreneurs / investors / PMs come to talk with her;inbound replaces inefficient networking(details in §4) |
| 3 | Friends | People who saw the content and identified with her views proactively ask for coffee → real friends. “What kind of energy you put out determines what kind of energy you attract.” |
| 4 | Cognition | Internal company blogs made colleagues willing to seek her out for conversations; experts she did not know well “were willing to talk after seeing my content” → turn them into advisors |
| 5 | Users | Product user interviews + early seed users = came from her followers. Followers even proactively wrote feedback documents |
| 6 | Candidates | Hiring = attracting “people who have seen your content and identify with your views.” Quality is far higher than blind online recruiting |
Core insight:All six returns are “content = trust, trust = skipping the cost of cold-starting with strangers.”.
4. Scaling relationships (Inbound > Outbound)
4.1 SaaS model mapping
Zara’s insight in the enterprise services industry: SaaS sales has two models → inbound vs outbound → directly maps to networking.
| Outbound (go find others) | Inbound (let others find you) |
|---|---|
| LinkedIn / email cold outreach | Writing content / making podcasts |
| Going to networking events | Hosting your own events |
| Most people don’t reply | People who identify with your views after reading come proactively |
| Most conversations are ineffective | People who come are high quality,self-filtered |
4.2 Two major inbound methods (Zara’s actual practice)
Method 1: Make content
“Almost all the connections I’ve accumulated over the past 10 years came through content.”
Mechanism:
- Write articles / make podcasts → others identify with them after reading → “let’s grab coffee and chat”
- Self-selection → people who reach out are high quality (identify with views / similar temperament)
- Especially suitable for introverts — reach tens of thousands of people without leaving home
Method 2: Gather people / host events(instead of attending other people’s events)
“Rather than attending 10 other people’s events, you might as well host one event yourself.”
Mechanism:
- You bear the venue / invitation / time / monetary costs → attendees are grateful to you → more willing to help you in the future
- Suitable for meeting second-degree connections: ask each friend to bring one of their friends
- vs. attending other people’s events → you can only talk to a few people + conversations don’t go deep
4.3 Real event: 100+ LinkedIn friend requests overnight
Zara wrote 2 English blog posts → posted them on LinkedIn → minor viral spread → 100+ friend requests overnight.
“Writing articles / making content is the most efficient way to meet like-minded people. It is 100 times more efficient than going to networking events and forcing awkward conversations.”
4.4 Luck = surface area × probability
Zara’s core argument on “how to become luckier in the workplace”:
Luck = surface area (your exposure) × probability (hit rate of a single event)
3 ways to expand your surface area:
- Create publicly: Learn in public, build in public, and naturally attract like-minded people
- Expand your circle:Weak ties(friends of friends / people connected through content) provide the most good opportunities
- Show up: Go to small events / post something / help a friend test a product — you never know which one will open a door
Every career turning point for Zara has been an application of this formula(VC / ByteDance / transition to AI).
“The more people who know how capable you are, the more likely it is that when an opportunity appears, someone will think of you immediately.”
5. How to persist (post more + stay mentally light + learn in public)
5.1 70,000 followers came down to two words:post more
"Good work, consistently, over a long period of time."
It is not about relying on luck to “make viral posts”; it is the compound interest of good × frequency × time compound interest.
5.2 Stay relaxed mentally (don’t be anxious about data)
“I dare say that among bloggers of a similar scale, I spend the least time on content creation and have the strongest sense of ease.”
| Anxious approach (❌) | Relaxed approach (Zara ✅) |
|---|---|
| Anxious about every post’s data | “Post casually, depending on mood” |
| Study viral-post formulas | Don’t study trending topics |
| Strict content planning | “Post whatever I feel like posting” |
| Must write a script | “Blurt it out” |
| Reflect whenever data is poor | Don’t reflect on individual posts; keep the rhythm |
5.3 Learn in public merges “output” and “learning”
“When I was learning AI, Xiaohongshu was basically my notebook. Whenever I gained insights from watching YouTube, I posted, achieving three things at once.”
1. Organize your own thinking through writing / output (a learning action you would need to do anyway)
2. Not only do you learn yourself, you also incidentally help others learn → build a personal brand
3. **Persona reinforcement effect**: unconsciously create a “passionate about learning AI” persona for yourself
→ In order to live up to this persona, you become more motivated to keep learning
The positive feedback loop of point 3 is the most important — learning in public is not just sharing,it strengthens your motivation to learn.
5.4 Build in public works the same way
Turn content output into a natural extension of work/life, so it can be sustainable.
Zara posted a retrospective video one day after launching Excalicord, and posted demo works while building Frontend Slides — product and content feed each other(details in Zara Zhang Zala / Skill & Independent Product Portfolio (Strategic Retrospective) §1).
“Building an audience and building a product feed each other bidirectionally.”
6. The content market is “made” (TAM grows 10×)
6.1 Counter-consensus insight
Zara initially thought: “For purely amateur, purely practical-knowledge sharing like mine, the TAM on Xiaohongshu is at most 10,000 followers.”
Reality:It grew to 100,000, and doubled in the most recent 2 months, driven by organic word-of-mouth recommendations.
“TAM is not figured out by thinking; it is made.”
Mechanism:The ceiling of a content product cannot be deduced through the logic of the ‘track’ it belongs to; it depends entirely on the specific form and execution.
Analogy: It is the same for products — “this track’s ceiling is only X hundred million” is a deduction out of thin air; in reality it depends on how far the people building it can push it.
6.2 Isomorphism between the content market and the product market
| Dimension | Making content | Making products |
|---|---|---|
| TAM | Determined by specific form + execution | Same |
| No shortcuts | Day-after-day investment | Same |
| Mindset | Stay relaxed | Same |
| Early stage | Compete on positioning | Same |
| Growth | Organic word-of-mouth recommendations | Organic word-of-mouth recommendations |
| Feedback | Comments / reposts | User interviews |
Details in Zara Zhang Zala / Skill & Independent Product Portfolio (strategy retrospective) §1 — the application of the same “TAM is made” to products.
7. Tool stack (minimalist)
| Stage | Tool | Alternative |
|---|---|---|
| Image-text creation | Xiaohongshu App | — |
| Video recording | Phone selfie / computer screen recording | No professional equipment / no lighting / no makeup |
| Video editing | Jianying AI subtitle editing | Done in 10 minutes |
| PPT presentations | Frontend Slides skill(own product) | Replaces PowerPoint |
| Screen recording tool | Excalicord(own product) | Replaces paid screen-recording software |
| Blog layout | OpenClaw + Feishu | — |
| Podcast recording | iPhone built-in recorder + one trick | Details in post 047, “How to Record a High-Quality Podcast with Your Phone” |
| Image generation | GPT (directly generates accompanying images) | — |
| Idea capture | iOS Notes / Drafts | Usually toss inspirations in casually |
Counter-consensus:No paid tool is necessary. Zara uses paid tools to save time (editing tools), not to improve quality.
8. Division of labor among 4 platforms in the content ecosystem
Zara’s platform matrix / each platform’s role:
| Platform | Followers | Content format | Audience | Purpose |
|---|---|---|---|---|
| Xiaohongshu | 260,000+ | Image-text posts + short videos (Chinese) | Chinese PMs / liberal arts students / job seekers | Main battlefield, main stage for build in public |
| 50,000+ | Short tweets (Chinese and English) | International AI builders | Source for finding AI technical ideas / international exposure | |
| 14,000 | English long-form posts | North American professional network | Job search / work opportunities | |
| Internal company blog | 24,000 readers | Chinese long-form posts | ByteDance colleagues | Internal influence / recruiting |
Key strategy:Twitter is a source of technical ideas (watching builders), Xiaohongshu is for Chinese output (influencing Chinese PMs/liberal arts students), and LinkedIn is for work opportunities. Content is not duplicated; each platform has its own content format.
9. Comparison with other self-media creators
| Dimension | zara | newtype Lao Huang | dontbesilent | Wang Kai |
|---|---|---|---|---|
| Main battlefield | Xiaohongshu + Twitter | YouTube + Substack | Xiaohongshu + X | Knowledge Planet + overseas platforms |
| Follower scale | 260,000 + 50,000 | Medium | Similar | Planet paid users |
| Style | Relaxed + opinionated | Engineered + extreme SOP | AI self-media SOP | Full-platform matrix |
| Rhythm | 1 post every 2 days, 4 years | Weekly YouTube + tweets | Medium frequency | Multiple posts daily |
| Monetization | Zero ads | Paid newsletter + skill | Selling content | Knowledge Planet + going overseas |
| Signature | Frontend Slides + 260,000 followers | Super Analyst three-piece set | Multiple skill bundles | tokrepo + FastFacts |
Details in AI Self-Media and Content Engineering (Cross-Author Thematic Synthesis) cross-author thematic summary.
10. Overview: Zara’s content creation philosophy
Inbound > Outbound
↓
Learn in public
↓
┌──────────────────────┐
│ Content is an interest, not work │
│ ≤ 30 min/post, posted casually │
└──────────────────────┘
↓
Persist (1 post every 2 days × 4 years)
↓
┌────────────────────────┐
│ surface area × probability = opportunities │
│ (jobs / friends / users / recruiting)│
└────────────────────────┘
↓
No ads / not doing it full-time
(avoid having the content format hijacked by advertisers)
↓
Content ↔ product feed each other bidirectionally
Related wiki
- Zara Zhang Zala / Vibe Coding Mental Model System — Zara’s vibe coding mental model system
- Zara Zhang Zala / Skill & Independent Product Portfolio (Strategy Review) — Zara’s product portfolio + build in public connection
- Zara Zhang Zala / follow-builders skill and AI Information Source Curation Methodology — follow-builders skill (Zara’s information-source management practice)
- AI Self-Media and Content Engineering (Cross-Author Thematic Synthesis) — Cross-author summary of AI self-media methodologies
- Zara Zhang Zala / North American Career Development SOP (Liberal Arts Student → VC → ByteDance → AI Independent Developer) — North American job search (planned) — strongly connected to how content brings work opportunities
History
- 2026-05-22 — Created based on a synthesized rewrite of 62 posts in the 02_content creation category (20+ high-density long descriptions) plus video posts with transcripts such as 047/049/006. 10 core mental models + numerical confidence (260,000 followers / 50,000 followers / 2 years / single post ≤30 min)
2026-05-31 supplementary note: boundaries of the self-reported 300,000 followers and the old tool list
In this full written note, the author states that she surpassed 300,000 followers, began posting relatively regularly about three years ago, has a full-time job, and spends no more than five hours per week creating content. She takes no commercial orders, does not chase trending topics, and does not produce polished visuals; when she learns something in daily life or tries a product she likes, she shares it along the way. She emphasizes that the most valuable gains are making friends and building a community (a group formed around shared interests), and invites readers to share their specific experiences.
This adds a dated personal case to the earlier point about “lowering the barrier to output”; it is not a causal promise that spending five hours a week will lead to growing to 300,000 followers. She also explicitly says that the content creation described in this self-account “does not use any AI”; therefore, the AI subtitles, images, and other tools listed in the earlier tool stack cannot be interpreted as her unified production workflow at this point in time. The two sources retain their own dates and wording, and no explanation for the discrepancy is offered on the author’s behalf. What amateur creators can take away is turning a genuine learning process into low-burden output; if the goal is commercial delivery by a team, review, pacing, and quality standards must be designed separately, and her personal unplanned approach cannot simply be copied.
This section is based on the author’s main text; this batch of comments from real people has now been read, and the boundaries of the feedback are noted in the supplement at the end. Comment stories cannot be used to estimate the general effectiveness of the content. The images have not been visually reviewed.
April–May 2026 supplement: Demo Day format and feedback boundaries
The April 17 event text provides a reusable format: invite 10–20 people who use agents deeply to screen-share and demonstrate their daily work; auditing requires no screening, while demonstrating requires an application. The author’s April 18 comment further specifies about 5 minutes per person, priority for small teams, and that what is shown is a workflow rather than product promotion. It is suitable for turning the abstract idea of “using AI well” into observable work actions; this is the design of a historical event, not a still-open invitation to register.
Three replay transcripts from April 27–29 provide viewing guides: the first covers a cognitive/trading assistant, work retrospectives, AI cowork, and Moxt; the second covers an external brain, an 8-person and 66-agent team, Claude Code and OpenClaw, a Feishu workspace, and multi-agent systems; the third covers video work, Scrum, travel planning, knowledge management, and career mentorship. The May 19 description of Ren Chuan’s sharing lists the division of labor among 5 agents, why he only chats with one, and time allocation for a one-person company.These are topics in the program descriptions, not implementation plans that have been read; team sizes and agent counts are also only figures from the program introductions and cannot be used to infer production efficiency. The videos still need to be transcribed.
On April 22 and May 6, the author emphasized the event as a method of collective learning; this connects with this page’s advocacy of learning in public. In the May 31 comments from real people, some said they started creating, built tools, and found like-minded people; at the same time, some were hindered by installation, costs, environment, and knowledge barriers. The comments are voluntary, unverified self-reports and lack an overall denominator of those who did not succeed; a few follower-growth or revenue stories cannot be treated as the method’s success rate.
This batch of texts and comments from real people has been organized; the images have not yet been visually reviewed. Recruitment images, registration methods, expired meeting numbers, and generic praise remain in the original draft and are not converted into current operational instructions. Robot summaries such as “Wen Yi Wen” are still not used as a basis for video content.
Sources and related materials
- https://www.xiaohongshu.com/explore/6a0b6eae000000003601db31
- https://www.xiaohongshu.com/explore/69efb709000000003703436a
- https://www.xiaohongshu.com/explore/69eefa6d000000003700eade
- https://www.xiaohongshu.com/explore/69e8618a000000002301475b
- https://www.xiaohongshu.com/explore/69fa1dd900000000230067dc
- https://www.xiaohongshu.com/explore/69f1a54500000000360023e7
- https://www.xiaohongshu.com/explore/69e17ae8000000002003ba1f
- https://www.xiaohongshu.com/explore/6a1b99de0000000007029f4f
- Zara Zhang Zala / Vibe Coding mental-model system
- Zara Zhang Zala / Skills & indie product portfolio (strategy retrospective)
- AI self-media and content engineering (cross-author thematic synthesis)