Case Study: SOI MOI Tarot — AI-powered Telegram Bot for Predictions

# Case Study: SOI MOI Tarot — AI-powered Telegram Bot for Predictions

Project Overview

SOI MOI Tarot is an innovative Telegram bot that combines ancient Tarot card traditions with modern artificial intelligence capabilities.

**Bot Link:** [tarot.soi.moi](https://tarot.soi.moi)

Client Challenges

Problems to solve: 1. **Create unique product** in Tarot niche 2. **Provide personalized predictions** for each user 3. **Scale business** without hiring more fortune tellers 4. **Create engaging UX** for user retention

Target Audience: - People interested in esoteric and self-knowledge - Age: 25-45 years - Predominantly female audience (70%) - Geography: Russia, CIS, Russian speakers

Technical Solution

System Architecture:

[Telegram API] ↔ [Bot Handler] ↔ [AI Engine] ↔ [Tarot Database]
                                      ↓
                               [User Analytics]

Technologies Used: - **Backend:** Python 3.11, aiogram 3.0 - **AI:** OpenAI GPT-4, custom prompting - **Database:** PostgreSQL with Redis caching - **Deploy:** Docker, DigitalOcean Droplets - **Analytics:** Yandex Metrika, custom metrics system

Key Features:

#### 1. AI-powered Card Interpretation
```python
async def interpret_cards(cards, question, user_context):
prompt = f"""
You are an experienced Tarot reader. User asked: "{question}"

Cards drawn: {cards}

Provide detailed interpretation considering:
- Meaning of each card in context
- Card interactions
- Practical advice

Style: wise but accessible, with mystical elements
"""

response = await openai_client.chat.completions.create(
model="gpt-4",
messages=[{"role": "system", "content": prompt}],
max_tokens=1000,
temperature=0.8
)

return response.choices[0].message.content
```

Development Process

Phase 1: Research and Prototype (1 week) - Competitor analysis - Tarot traditions study - Basic bot prototype creation

Phase 2: MVP Development (3 weeks) - Basic architecture creation - OpenAI API integration - Main spreads implementation - Focus group testing

Phase 3: Improvements and Scaling (2 weeks) - Personalization addition - Performance optimization - Interface A/B testing - Analytics implementation

Project Results

Success Metrics:

User Reviews:

> *"Simply incredible! Predictions are so accurate, as if the bot knows me personally. Use it every day!"*
> — Anna, 32 years old

> *"Card interpretation quality is on par with professional fortune tellers. Very convenient to use in Telegram."*
> — Mikhail, 28 years old

Technical Challenges and Solutions

1. AI Prediction Quality **Problem:** Generic, impersonal responses from GPT

2. Performance Under High Load **Problem:** Response delays with user growth

3. User Engagement **Problem:** Low user return after first use

Lessons and Insights

What worked well: 1. **AI + traditions** — combining modern tech with ancient practices 2. **Personalization** — users value individual approach 3. **Ease of use** — intuitive interface 4. **Regular content** — daily predictions create habit

What can be improved: 1. **More spreads** — expand reading variety 2. **Social features** — share predictions with friends 3. **Education** — add Tarot learning section 4. **Integrations** — connect astrological data

ROI and Business Metrics

Investment: - Development: $8,000 - Infrastructure: $200/month - AI API: $350/month - **Total costs:** $8,000 + $550/month

Revenue: - Month 1-3: $1,200/month - Month 4-6: $2,800/month - Month 7-12: $4,200/month

ROI: 425% in first year

Development Plans

Short-term (3 months): - New spreads addition - Astrological data integration - A/B test new notification types

Long-term (6-12 months): - Web version of bot - Mobile application - International localization - Partner program

Conclusion

SOI MOI Tarot project demonstrates how proper AI integration can create a unique product in traditional niche. Key to success — balance between technological capabilities and user needs.

**Want to create similar project?** Contact us: [@mixvlad](https://t.me/mixvlad)