💡 Inspiration
In today’s fast-paced digital world, online conversations shape relationships, careers, and communities. Yet many users struggle to express themselves clearly or strategically—especially across cultures, platforms, and social roles. We wanted to build a tool that empowers people to communicate with empathy and precision, using AI not just to rewrite words, but to guide relationships. Our goal: turn everyday chats into intentional, trust-building interactions.
🛠️ What it does
This Chrome extension acts as a real-time AI companion layered over social media chat interfaces. It supports two modes:
Single Chat Mode:
Users input their social intent (e.g., “be friendly,” “set boundaries”)
AI rewrites messages based on full chat history and user goals
Provides a structured chat plan with timeline-based suggestions for building rapport
Group Chat Mode:
Includes all Single Chat features
Analyzes each participant’s personality and role based on their messages
Offers tailored strategies for navigating group dynamics
The interface is a floating, resizable window that adapts to the user’s typing language and supports platforms like WeChat, WhatsApp, and Messenger.
🧱 How we built it
We used:
JavaScript + Chrome Extension APIs for browser integration
OpenAI API for natural language understanding and rewriting
LocalStorage and IndexedDB for chat history caching
CSS Grid + Flexbox for dynamic UI resizing
Language detection via Intl and custom heuristics for multilingual support
We also designed a modular architecture to support future platform integrations and privacy-safe local inference.
🚧 Challenges we ran into
Ensuring chat history context was preserved across sessions
Designing a UI that’s both unobtrusive and informative
Balancing real-time performance with deep language analysis
Handling multilingual input and output without latency
Inferring personality traits from limited group chat data
🏆 Accomplishments that we're proud of
Built a working prototype that rewrites messages based on full chat context
Designed a flexible UI that adapts to user intent and screen size
Created a relationship-building planner that feels intuitive and actionable
Enabled multilingual support across major platforms
📚 What we learned
Real-time AI suggestions require thoughtful UX to avoid cognitive overload
Users value not just better wording, but strategic guidance
Language models can infer social roles surprisingly well—but need guardrails
Privacy and transparency are key when working with personal conversations
🚀 What's next for Real-Time AI Companion for Social Media Chats
Integrate with more platforms (Telegram, Discord, Slack)
Add emotion detection and tone calibration
Enable local AI inference for privacy-first deployments
Launch a mobile version with gesture-based input
Partner with mental health and communication coaches to refine chat plans
Log in or sign up for Devpost to join the conversation.