24quan \China

Step 2 - AI-Assisted Hybrid (Product-Market Fit): Replace 50% of manual curation with LLM-powered recommendations. Fine-tune GPT-4 on 10,000+ Shanghai restaurant reviews (scraped from Dianping, Xiaohongshu, translated Michelin guides) and merchant metadata. Build a RAG (Retrieval-Augmented Generation) pipeline: user query -> embed with OpenAI -> semantic search in Pinecone for top 20 merchants -> LLM re-ranks based on context (date, party size, dietary restrictions, past preferences) -> return top 3 with explanations. Human concierges QA every AI response before sending. Expand to 500 merchants across Shanghai, add 5 new categories (spas, private tours, yacht rentals, art galleries, personal trainers). Launch referral program: users who invite 3 friends get 1 month free. Goal: 1,000 paying subscribers, 80% AI-assisted queries, 30% month-over-month growth, $50K MRR ($30K subscriptions + $20K commissions).

Step 3 - Merchant SaaS Platform (Moat Building): Build a beautiful merchant dashboard (Next.js + Supabase) that makes Qingke indispensable to supply side. Features: real-time reservation calendar synced with WeChat, customer CRM with preferences and visit history, AI-powered yield management (suggest optimal pricing for slow periods), automated review requests post-visit, and analytics on customer acquisition sources. Charge merchants $200/month SaaS fee (in addition to 10% booking commission) but waive it for first 6 months to drive adoption. This creates switching costs—once a merchant's operations run on Qingke, moving to Meituan means losing customer data and workflow integrations. Expand to Beijing and Shenzhen, targeting 2,000 merchants and 5,000 subscribers. Add premium tier ($50/month) with perks like priority booking, exclusive merchant access, and personal account manager. Goal: $200K MRR, 40% from SaaS, 60% from commissions and subscriptions.

Step 4 - Data Moat and Vertical Expansion (Scale): By now, Qingke has proprietary data on affluent consumer preferences that Meituan lacks—what time they book, how they describe experiences, which merchants they return to. Use this to build a recommendation engine that's genuinely better than Douyin's algorithm for premium experiences. Launch 'Qingke for Business'—corporate concierge for client entertainment (huge market in China, currently handled by assistants making calls). Partner with Amex China, China Merchants Bank, and luxury hotel chains to offer Qingke as a cardholder perk. Expand into adjacent verticals: wellness (TCM doctors, personal trainers), travel (boutique hotels, private guides), and education (music teachers, language tutors). Each vertical uses the same AI infrastructure but different merchant onboarding playbooks. Goal: $1M MRR, 20,000 subscribers, 10,000 merchants, Series A fundraise ($10M at $50M valuation) to expand to 15 cities and build out the team.