Skip to content

README.md

Latest commit

 

History

History
108 lines (77 loc) · 4.75 KB

README.md

File metadata and controls

108 lines (77 loc) · 4.75 KB

Athena: Distributed AI System + Collaboration Research

Research through practice: Building a real distributed AI system while developing universal AI-human collaboration methodology

What This Is

Athena is a distributed AI system with physical components (cameras, sensors, displays) that make intelligent decisions and control devices. What makes it unique is that it's built using research through practice - simultaneously developing the system AND the methodology for effective AI-human collaboration.

🎯 Dual Purpose Design

Practical Output: Distributed AI system with:

  • Physical components - PTZ cameras, sensors, displays, Raspberry Pis
  • Event processing - Motion, sound, MQTT, file changes, Home Assistant integration
  • AI decision making - Process events, control devices, display information
  • Real-time interfaces - Phoenix LiveView dashboards and chat system

Research Output: Universal AI-human collaboration patterns:

  • Physics of Work methodology - Make it better for the next person (which is us)
  • Two-role framework - AI as development team, human as product team
  • Conversation archaeology - Complete development consciousness preservation
  • Decision confidence protocols - Systematic risk assessment for AI autonomy

🔍 What Makes This Unique

This isn't just building software OR studying methodology - it's research through practice. Every component is both functional system code AND a laboratory for discovering collaboration patterns that work at scale.

Quick Start

Running the System

# Start the chat interface (main application)
cd ash_chat && mix phx.server

# Access the interface
open http://localhost:4000

Demo the Multi-User Chat

# Start an IEx session
cd ash_chat && iex -S mix

# Create demo data with users, rooms, and AI agents
AshChat.Setup.reset_demo_data()

Architecture Overview

  • ash_chat/ - Production Phoenix LiveView chat system with AI agents
  • chat-history/ - 90 conversations (50MB) of complete development archaeology
  • docs/journal/ - 30+ real-time development discoveries
  • domains/ - Event processing, hardware integration (migration in progress)
  • system/ - Hardware controls, computer vision, MCP servers

Core Components

Multi-User AI Chat (ash_chat/)

  • Agent Cards - Character-driven AI personalities
  • Room Hierarchy - Parent-child room relationships
  • Real-time Events - LiveView interfaces with auto-discovery
  • Multi-model Support - Switch AI providers per conversation

Conversation Archaeology (chat-history/)

  • 90 conversation files with structured JSONL metadata
  • Complete development timeline from project inception
  • Tool usage, decisions, and context fully preserved
  • Git correlation linking conversations to specific commits

Hardware Integration (system/)

  • PTZ camera controls via MCP servers
  • Computer vision pipeline (SAM, CLIP)
  • Event collection from multiple sources
  • MQTT/sensor integration

Research Value

Physics of Work in Practice: This project demonstrates:

  • Iterative process improvement - Each component development improves the methodology
  • Knowledge transfer patterns - AI learns from human expertise, human learns from AI systematic thinking
  • Decision confidence protocols - Proven risk assessment for autonomous AI development
  • Universal applicability - Patterns that work across project types and technical domains

Documentation

Essential Reading

For AI Collaborators

Impact

Practical: A working distributed AI system with real hardware integration, event processing, and intelligent decision-making capabilities.

Methodological: Proven patterns for AI-human collaboration that scale beyond individual projects, with complete development consciousness preservation as a side benefit.

Future: Universal methodology that any AI-human team can adopt for effective collaboration, regardless of domain or technical stack.

Research through practice: Building the future while learning how to build it better.