A Smart AI-powered Social Listening & Simulation Platform for Brands, Creators, and Analysts
init0-sentinel is a multi-platform AI-powered post analysis and audience simulation tool built to help:
- Brands 📢
- Influencers 🎥
- Researchers 📊
- Everyday users 👥
...to understand how their post will perform before actually posting on major platforms like Instagram, Twitter (X), Threads, LinkedIn, etc.
- Users log in via Google Authentication
- Enter post content, attach image(s), and select platform(s)
- Image is uploaded to Cloudinary
- Post metadata is saved in Firestore
- Once uploaded, users can navigate to the History page
- There, users view analysis cards with post metadata and a "Get Full Insights" option
While our current implementation uses Azure OpenAI (GPT) and Google Gemini models due to time constraints during Hack8, the system is fully modular — allowing integration with:
- 🔁 Your own fine-tuned ML models
- 🧠 HuggingFace Transformers
- 📦 Local LLMs / REST APIs / Python-based sentiment classifiers
⚙️ The key strength lies in our high-level prompt engineering:
Not just asking vague questions — every prompt is structured to extract exact information like:
- Engagement breakdown by age group
- Seasonality relevance
- Hashtag trends
- Platform-specific response patterns
Outputs are well-structured, JSON-based, and ready for further analysis or visualization
When analyzed, each post gives deep insights including:
- Audience Segmentation
- Sentiment Summary (Radar Chart)
- Public Reactions
- Platform-specific Behavior
- Hashtag Popularity
- Strategic Recommendations
... all derived using AI and web-scraped real-world reactions.
All backend logic is handled under routes/index.js and flows as:
- 🔍 Post content sent to Azure OpenAI → Extracts keywords & tones
- 🌐 Keywords passed to Gemini Grounding Search → Web-scrapes user reactions, trending hashtags, etc.
- 🤖 Scraped data returned to OpenAI → Generates:
- Audience breakdown
- Engagement strategies
- Sentiment analysis
- Radar chart scores
To avoid relying on external paid or closed APIs (like Instagram API), we built our own PWA social media platform to simulate audience engagement.
- Users can post content
- Others can:
- ❤️ Like
- 💬 Comment
- 🎭 React with emojis
- Post reactions feed into backend to refine future predictions
This provides pre-launch simulation of how a post might perform in real-world social media.
init0-sentinel/
├── Desktop-Client # Electron or local desktop dashboard
├── PWA-Client # PWA-based social simulation app
├── backend # Node.js server with OpenAI/Gemini integration
├── LICENSE
├── README.md- Frontend: React (ShadCN, Tailwind), Vite
- PWA: React PWA App with Firebase
- Backend: Node.js, Express
- Database: Firebase Firestore
- AI Models: Azure OpenAI, Google Gemini (pluggable)
- Image Hosting: Cloudinary
Built during Hack8 by a passionate group of developers who believe in intelligent, data-backed storytelling.