The archive · AI & Models · Product decision · 2024–2026
NoonWake bets AI fortune companions that remember you win on retention
Ex-Xiaomi director Shawn Wu's fortune-AI app Wanxiang Youling hit ~500k users, ~40k DAU and ¥300k+/month revenue before a ~¥10M seed.
NoonWake.AI (Wanxiang Youling / 万象有灵)
What the business is
NoonWake.AI builds AI fortune-telling and emotional-companion apps — Wanxiang Youling (万象有灵) in China and Starot abroad — plus a desk hardware device, pairing a fine-tuned Qwen model with the Chinese yin-yang 'Ten Gods' personality framework.
Starting capital：Nearly ¥10M (RMB) seed round announced June 2026, led by Yunqi Capital with GeekPark's Variable Capital as follower.
How it started
Wu (b. 1994) moved from ZTE to Alibaba's City Brain team to Tencent, where he co-built the social app Island (¥30M+ pre-A) and watched it die; at Xiaomi he became one of the youngest ecosystem directors. He left in 2024, spent nearly a year searching, and concluded that in the AI era the moat is whoever keeps receiving users' real, ongoing context — which emotional and spiritual conversations extract naturally.
What happened
Wanxiang Youling launched in Oct 2025 with no ad spend and passed ~500k users and ~40k DAU by June 2026. Yunqi Capital and GeekPark's Variable Capital led a nearly ¥10M seed announced 2026-06-17. For overseas markets the team built Starot (tarot/astrology, $9.9–15.9/month, 30k+ pre-registered before launch; live on Google Play by Sep 2026) and planned Arday for Christian users. Its 'Lucky Calendar Machine' desk device — a 48-hour prototype that won second prize in Xiaohongshu's 2026 hackathon hardware track — was slated for a Q3 2026 Kickstarter at $199–259.
How it ended up
Still live and expanding as of 2026-09-02: Wanxiang Youling passes 400k users in China, Starot is live on Google Play, and the company is pushing the companion idea into hardware; the open question is whether paying users stay beyond the novelty of fortune content.
Background
NoonWake.AI is a Shenzhen/Hong Kong company founded by Shawn Wu, an ex-Tencent and ex-Xiaomi product director, that sells AI fortune-telling and emotional companionship. Its China app Wanxiang Youling turns the traditional yin-yang 'Ten Gods' personality framework into ten AI personas, runs on a fine-tuned Qwen model trained on roughly 800k real consultation transcripts, and remembers what users say so characters can follow up days later.
The founding bet was that the AI-era moat is continuous access to users' real context, and that spiritual and emotional conversations are where people expose it most willingly. Wu left a director post at Xiaomi in 2024 after an earlier startup failure at Tencent (the social app Island), and launched Wanxiang Youling in October 2025 with no paid acquisition.
By June 2026 the app had roughly 500k users and 40k daily actives, monthly revenue above ¥300k and subscription ARR in the low millions of RMB, with 60–70% of growth organic; Yunqi Capital and GeekPark's Variable Capital led a nearly ¥10M seed. The team then took the model overseas as Starot (30k+ pre-registered, $9.9–15.9/month) and planned a Christian companion app, while its Xiaohongshu-hackathon 'Lucky Calendar Machine' desk device headed to Kickstarter at $199–259.
As of 2026-09-02 the company is live and expanding: its site claims 400,000+ China users and Starot is downloadable on Google Play. The unproven part of the bet is retention — whether paying users treat the personas as long-term companions rather than a novelty to outgrow.
What has to be true
- Fortune-telling is a cheap, self-spreading acquisition wedge: users share horoscopes and daily readings without admitting they need emotional support.
- Long-term memory and proactive follow-up turn one-shot chats into recurring habits, the retention layer most fortune AI clones lack.
- Emotional and spiritual conversations make users expose real, ongoing life context — the asset Wu argues decides who wins the AI era.
- A desk hardware device keeps the product physically present and adds a daily context channel software alone cannot own.
- Separate localized products (Starot in North America, Arday for Christians) attack markets with mature subscription habits instead of China's weak willingness to pay.
What can be applied
Pick an entry point users will share (fortune), then a reason they stay (memory, proactive care); cheap acquisition and real retention are two different products.
Aftermath
As of 2026-09-02 NoonWake.AI is live and growing: Wanxiang Youling passes 400k users in China (DAU ~40k reported June 2026), Starot is live on Google Play, and the Lucky Calendar Machine heads to a Q3 2026 Kickstarter at $199–259 after a ~¥10M Yunqi-led seed. Risks: app-store policy volatility around mysticism content and competition from AI fortune and companion rivals entering the same markets.
Sources
- 他不做大厂总监，All in AI陪伴，月入数十万 (He quit his big-tech director job to go all-in on AI companionship, earning hundreds of thousands a month)
- “NoonWake.AI” has completed a seed round financing of nearly 10 million yuan
- 深圳一家初创企业用“玄学+心理学”重构人格化陪伴，完成近千万元种子轮
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