talk about it

• GIF

  demo of the app

• 

  image of the interface

About

talk about it is an asynchronous Python-based message processing system that bridges Apple's iMessage ecosystem with advanced AI models. Built on asyncio and leveraging native macOS automation, talk about it enables intelligent, context-aware conversations at scale.

Key Features

• Concurrent Processing: Handles $n \leq 20$ simultaneous conversations with $O(1)$ latency overhead
• Natural Interaction: Multi-message responses split on \n\n delimiters with dynamic typing delays $\Delta t \in [1.0, 3.0]$ seconds based on message length
• Smart Debouncing: Processes only the most recent message when users send rapid sequences, with configurable $\tau = 0.3s$ debounce period
• Persistent Memory: SQLite-backed conversation history maintains context across sessions
• Real-time Indicators: Typing bubbles and read receipts via GUI automation

Architecture

Message Detection → SQLite Polling (500ms)
AI Processing → Gemini API (concurrent, n=20)
Response Delivery → AppleScript Automation (serialized)

Technical Stack

• Runtime: Python 3.8+ with asyncio event loop
• AI Backend: Google Gemini 2.5 Flash via official SDK
• Message I/O: Direct SQLite access (chat.db) + AppleScript automation
• Concurrency Model: Semaphore-limited AI calls with GUI lock for serialization

Performance Metrics

Use Cases

• Therapy Assistance: 24/7 mental health support via familiar messaging interface
• Customer Service: Automated support through business iMessage numbers
• Personal AI Assistant: Contextualized help accessible through text messages
• Educational Tutoring: On-demand learning support via SMS/iMessage

Git repo: https://github.com/CaptainATW/hack-coms-therapy

Built With

• gemini
• python
• sql