← Sticker Wall AGENT · DETAIL

Breakup Shelter AI Healing Mini-Program: Kubler-Ross 5-Stage Companion & Crisis Referral with Monthly Revenue of 12K RMB

Workflow: Daily operations workflow: In the morning, update the latest emotional keywords for breakup topics on Xiaohongshu and We

AGENT

Key Fields

FIELD STAMPS

IndustryHealthcare / Elderly Care

RegionChina（国内）

ScaleSME

ChannelOnline

🔧 Workflow

Daily operations workflow: In the morning, update the latest emotional keywords for breakup topics on Xiaohongshu and Weibo to iterate the script library. When users enter the mini-program, they first complete a 1-minute emotional scale rating; subsequently, the AI recognizes emotional fluctuations in the dialogue in real-time, matching adaptive scripts and healing tasks across the Kubler-Ross five stages (Denial/Anger/Bargaining/Depression/Acceptance). In the evening, automatically generate the user's daily emotional trend report. If self-harm or suicidal keywords are identified, immediately trigger a three-level response: push the official crisis hotline, have manual customer service intervene within 10 minutes, and simultaneously generate a referral ticket dispatched to contracted counselors. Inputs include user emotional text, diary content, and initial scale scores; outputs include stage-adaptive dialogue responses, emotional trend reports, and crisis referral notifications.

🛠 Setup Requirements

Setup requires basic WeChat mini-program development capabilities (can be outsourced if without a background, cycle 2 weeks). Core tasks include debugging prompt words for the Kubler-Ross five-stage judgment and emotional scale scoring logic, using Coze to orchestrate dialogue workflows, and integrating the DeepSeek-V3 emotion recognition API without the need to independently train models. It is necessary to contract 3+ psychological counselors holding trauma healing qualifications in advance to handle referral tickets, purchase platform liability insurance, and complete personal information protection compliance filings. For individual developers, it takes about 4-6 weeks from demand confirmation to MVP launch.

🧰 Toolchain

🔧 Coze
🔧 WeChat Mini-Program
🔧 DeepSeek API
🔧 Kubler-Ross Grief Scale
🔧 Lark Bitable

💰 Revenue

Monthly revenue is about 12,000 RMB, consisting of: 1. Membership subscription fees: 500 daily active users calculated at an 8% paid conversion rate, 40 paying members × 29 RMB/month = 1,160 RMB; 2. Deep healing packages: 30 orders/month × 99 RMB/order = 2,970 RMB, including AI-customized healing plans + 15 minutes of 1v1 communication with a counselor; 3. B-end university partnerships: providing mental health supporting services to 2-3 universities, with a single school monthly service fee of 3,000 RMB, totaling about 7,870 RMB. Combined with minor advertising revenue, total monthly revenue is stable at around 12,000 RMB.

💸 Cost

Monthly cost is about 1,200-1,500 RMB, among which companion dialogue inference expenses (at the 500 daily active users level) are 300-500 RMB/month, WeChat mini-program authentication and cloud server fees are 200 RMB/month, platform liability insurance is 150 RMB/month, and psychological counselor commissions are about 30% of revenue, settled monthly.

⏱ Time Investment

Daily investment of 2-3 hours: 1 hour handling crisis tickets, user feedback, and content moderation; 1 hour updating operational content and the script library; half an hour analyzing user emotional data to adjust strategies. Spend 3 hours every Sunday holding a script calibration meeting with contracted counselors to synchronize the follow-up status of high-risk users.

🚀 Getting Started

The first step is to build a Minimum Viable Product (MVP): use Coze to quickly build a breakup listening demo, publish free trial notes in tree-hole and diary formats in Xiaohongshu and Douban breakup groups, and recruit 100 seed users for testing. Focus on tracking next-day retention rate (target ≥55%), user proactive re-chat rate, and emotional scale score changes. After validating the demand, launch the mini-program to enable payments, and simultaneously contact local psychological counseling institutions to discuss referral cooperation.

🔑 Keys to Success

✅ The Kubler-Ross five-stage script system needs to be calibrated and reviewed quarterly by licensed psychologists to prevent AI from outputting content that does not comply with psychological principles, which could cause secondary emotional harm to users.
✅ The crisis referral process must clearly define responsibility boundaries: AI only performs preliminary screening and ticket dispatching without bearing diagnosis and treatment responsibilities. All referral tickets must be undertaken by licensed counselors with service records retained.
✅ Traffic generation content should focus on authentic tree-hole and diary formats, strictly prohibiting false propaganda and anxiety marketing to avoid platform traffic throttling and compliance risks.
✅ Sensitive data such as user emotions and diaries must be stored with strict encryption, complying with the medical and health data regulations of the Personal Information Protection Law to prevent privacy leaks from causing compliance issues.
✅ The traffic acquisition cost for breakup topics on Xiaohongshu should be controlled within 3 RMB per user, focusing primarily on natural seeding to avoid low ROI from paid advertising.

⚠️ 风险

⚠️ Extreme emotional users engaging in self-harm or suicidal behavior, with family members holding the platform accountable for inadequate supervision, leading to legal disputes and compensation risks.
⚠️ Inadequately tuned AI scripts causing secondary emotional harm to users, leading to a collapse in reputation and massive user churn.
⚠️ Compliance issues regarding the qualifications of contracted psychological counselors. If counselors practice without a license or provide improper service, the platform must bear joint and several liability.
⚠️ Because the mini-program content involves the field of psychological services, it may trigger failed compliance reviews on the WeChat platform, resulting in product delisting and rectification.

📌 Real Cases

📌 TRAE competition project "Ferry" is the first AI breakup healing companion based on the Kubler-Ross grief curve. Through a stage-adaptive system, it accompanies users through the entire healing process from acute stage → processing stage → relief stage → reconstruction stage. Within 3 months of launch, it accumulated 23,000 users, validating the real demand in the segmented breakup emotion track.
📌 AI emotional companionship platform "XinLingZhiJi" (aisoul.xin) has continuously operated 24-hour AI emotional listening and crisis referral services, with monthly paid users stabilizing at over 8,000, proving the monetization path is viable.
📌 Segmented emotional companionship product "LianYu" focuses on segmented scenarios such as breakups and postgraduate entrance exam pressure. Within 6 months of launch, it accumulated 120,000 users with a paid conversion rate of 7.2%, and monthly revenue broke through 180,000 RMB.

Sources