The archive · Consumer Apps · Product decision · 2022–2026
CaiYan's AI 'guess the disease' game: Xiaohongshu jokes drive 1.6M users
Zhang Yue's quiz platform CaiYan went viral via its AI-patient game; 1.6M users, 200k peak DAU, no ads.
CaiYan (猜盐)
What the business is
CaiYan (猜盐), a pun meaning both 'guess the salt' and 'quiz', is a Chinese WeChat (微信) mini-program and web platform with dozens of AI information games — Guess the Disease, Guess the Country (猜国), Guess the Word (猜词), Guess the City (猜城), Guess the Song (猜歌), Guess the Cute (猜萌) — where players interrogate an AI patient or infer answers from clues, and daily challenges; it was incubated in 2022 by one-person-company founder Zhang Yue.
Starting capital：Bootstrapped: no investment raised as of 2026-04; the founder cites monetization from the launch year, 2023 revenue above ¥100k, and a ¥3M revenue target for 2026, with a team of one full-time founder plus two part-time students.
How it started
Zhang Yue, a computer-science graduate (two ACM Asia-regional bronzes, a national programming prize) who worked at Alibaba for years, first built the solo game Tuxun (图寻), which asks players to locate a place from street-view images and passed 1M+ users but was limited to geography. In October–December 2022 he incubated CaiYan to repeat the 'guess from information' loop across history, medicine, words, cities, songs and other domains.
What happened
CaiYan monetized from its launch year (2023 revenue above ¥100k) and kept adding modes. In May 2025 its Guess the Disease mode — the player is the doctor and an LLM plays a patient ordered not to reveal its illness — went viral as players shared interrogation screenshots on Xiaohongshu, turning diagnosis into an 'AI jailbreak contest'. Red Star News (红星新闻) found doctors and medical students using the game as a diagnosis drill, and the founder said its positioning was entertainment plus education. By the 2026-04 interview the platform had passed 1.6M cumulative users with a 200k DAU peak, over 30% next-day retention, revenue tripled in 2025, and a ¥3M target for 2026 — still run by one full-time founder, two part-time student helpers and about 10k core users in a QQ group.
How it ended up
Still live and profitable as of 2026-09-02: bootstrapped and solo-run; the founder presented at Chengdu's first OPC investor roadshow in March 2026 with talks underway but no round closed.
Background
CaiYan (猜盐) is a Chinese quiz-and-game platform built and run almost entirely by one person, Zhang Yue, a former Alibaba engineer. Its dozens of modes turn real-world information into guessing games — cities from images, songs from lyrics, countries, words — and its viral star is Guess the Disease, where the player acts as doctor and an AI plays a patient who must not reveal its own illness.
The story started with Tuxun, Zhang's GeoGuessr-like street-view location game, which passed 1M+ users but had a limited market. In late 2022 he incubated CaiYan to repeat the 'guess from information' loop across many domains; it monetized in its launch year, passing ¥100k in 2023 revenue before its 2025 viral wave.
In May 2025, Guess the Disease spread across Chinese social feeds as players posted absurd interrogation screenshots — PUA-ing the AI, jailbreaking its prompt, even convincing it to confess — which a popular tech account, Chaping, dubbed an 'AI jailbreak contest'. Red Star News reported doctors and medical students playing it as a diagnosis drill, and Zhang said the product aimed to combine entertainment and education.
By the China Business Journal interview of 2026-04-05 the platform had passed 1.6M cumulative users with zero paid acquisition, peaked at 200k DAU, typically held about 20k DAU with over 30% next-day retention, tripled revenue in 2025, and targeted ¥3M for 2026 — with one full-time founder, two part-time students and about 10k core users in QQ.
What has to be true
- Guess the Disease inverted a familiar format — the AI is the patient, the user is the doctor — giving players a novel stage for jokes and role-play.
- User-generated diagnosis posts carried the game to Xiaohongshu and beyond, making distribution free for a founder with no marketing budget.
- Tuxun's 1M+ users validated the 'guess from information' loop before CaiYan generalized it across dozens of domains.
- AI removed the content bottleneck, letting one founder operate a platform that would normally need a content team.
- Daily single-patient challenges and high difficulty created scarcity and bragging rights, driving repeat visits and shares.
What can be applied
The product is the ad: when a game gives users a stage for humor and self-expression, their posts do the distribution that a solo founder could never afford to buy.
Aftermath
As of 2026-09-02 CaiYan is still live and profitable, bootstrapped and solo-run: after 1.6M cumulative users, a 200k DAU peak and over 30% next-day retention by April 2026, Zhang targeted ¥3M revenue for the year with one full-time founder, two part-time students and a ~10k-user QQ community. He presented at Chengdu's first OPC investor roadshow in March 2026; talks began but no round closed, and he questions whether VC suits a small vertical solo product. Remaining risks: trend-driven modes, LLM updates absorbing features, and heavy competition.
Sources
- 探路“一人公司”｜对话“猜盐”创始人张月：OPC的“生死线”在于快速落地与商业化验证
- 爆火的“猜病”游戏里，集齐了全中国的电子庸医 (The viral 'guess the disease' game, full of China's electronic quacks)
- AI扮演病人的“猜病”游戏走红，已有多家医院引用AI标准化病人助力考核教学
spotted an error? The archive wants to know.
Your turn
You just read one. Describe what you are building, and see who is betting on the same thing.
Free account · 3 free questions · no card