← Sticker Wall AGENT · DETAIL

Breakup and Graduate Exam AI Listening & Crisis Referral System: A Secret Side Hustle Making 15,000 RMB a Month

Workflow: Operators first check the high-risk emotional warning dialogues pushed by the system every day, prioritize processing ri

AGENT

Key Fields

FIELD STAMPS

IndustryHealthcare / Elderly Care

RegionChina（中国大陆）

ScaleSME

ChannelOnline

🔧 Workflow

Operators first check the high-risk emotional warning dialogues pushed by the system every day, prioritize processing risk cases, and manually intervene in crisis intervention referrals to connect with professional psychological hotlines or counseling services. Afterwards, they review the emotional tags and reply quality of low-risk dialogues from the previous day, and iterate empathy prompt words. The system automatically processes hundreds of ordinary emotional venting entries daily. After inputting the text entered by the user, it first completes emotional stage classification and risk scoring. Low-risk queries output empathy replies adapted to the scenario, while high-risk queries trigger early warning pushes to the operator's end. The emotional tags of all conversations are automatically synchronized to the user database for long-term companionship strategy optimization.

🛠 Setup Requirements

Building the system requires no professional programming skills; mastering basic prompt engineering and low-code platform operations is sufficient. First, register a Coze or Dify account, configure emotional classification nodes (calling DeepSeek or other large model APIs to complete emotion recognition and risk scoring), build reply generation nodes, and pre-load empathy script libraries and crisis intervention referral rules for segmented scenarios such as graduate exams and breakups. At the same time, prepare a WeChat Mini Program or Official Account as the user entry point, configure the basic server and content moderation interface. The overall setup cycle takes about 1 to 2 weeks for online trial operation.

🧰 Toolchain

🔧 Coze
🔧 Dify
🔧 DeepSeek API
🔧 WeChat Mini Program

💰 Revenue

Monthly income is about 15,000 RMB, mainly coming from three parts: First, segmented scenario membership subscription fees, launching monthly/quarterly companionship memberships for graduate exam groups and breakup groups, with a unit price of 29-99 RMB/month. Accumulating 1,000 paying users can cover costs. Second, crisis intervention referral service fees; after connecting with professional psychological counseling institutions, successfully referring users can yield a referral commission of 100-300 RMB/time. Third, customized companionship service fees, launching personalized companionship plans for users with deep needs, with charges ranging from several hundred to over a thousand RMB.

💸 Cost

Monthly fixed costs are about 300-800 RMB. Among them, the basic version subscription fee for the low-code platform is about 100-300 RMB/month, large model API usage fees are billed based on consumption, averaging about 100-200 RMB/month, and WeChat Official Account certification fees and mini-program server deployment fees are about 100 RMB/month. In the initial stage, new user platform discounts can also be enjoyed, further compressing costs.

⏱ Time Investment

Investing about 2 hours a day, mainly used for manual intervention in handling high-risk emotional warnings, reviewing companionship dialogue quality to iterate prompt words, and user maintenance of private domain communities. Ordinary dialogue replies and emotional classification are automatically completed by AI without manual intervention.

🚀 Getting Started

The first step for beginners is to select 1 vertical segmented scenario, such as 'Anxiety Companionship in the Last 30 Days of Graduate Exams' or 'Emotional Guidance in the First Week After a Breakup', collect real emotional venting content published by users on social platforms in this scenario, and organize it into training corpus. Then, register a low-code platform account, build a dialogue workflow according to the preset three nodes of emotional classification, reply generation, and crisis warning. After testing and adjusting the accuracy of the scripts, publish it to platforms like WeChat Official Account or Xiaohongshu, accumulate private domain traffic by publishing emotional guidance dry-goods content, and recruit the first batch of 100 seed users for testing and iteration.

🔑 Keys to Success

✅ Accuracy of the crisis warning dictionary and the division of manual intervention boundaries
✅ Quality of empathy script prompts for segmented scenarios such as graduate exams and breakups
✅ Traffic acquisition and long-term repurchase maintenance of private domain communities
✅ Accurate mapping of anxiety sources for segmented crowds and scenario-based adaptation of scripts

⚠️ 风险

⚠️ If crisis warnings fail and lead to extreme events, it will face moral controversy and legal responsibility
⚠️ Long-term companionship can easily cause ethical risks of users developing false emotional dependence on artificial intelligence
⚠️ Platform content compliance risks; some social platforms have review restrictions on AI emotional companionship content, which may lead to accounts being banned

📌 Real Cases

📌 "Dukou", a breakup healing companion based on the Kubler-Ross grief curve, accompanies users through the closed loop from the acute phase to relief
📌 AI emotional chat companions use automated tools to provide companionship for single youth, realizing a side hustle monetization of over 10,000 RMB per month for a single person
📌 Xinyu, as an AI mental health companionship project targeting college students, has been implemented in university scenarios to provide round-the-clock emotional venting services

Sources