pAI – Personal AI Systems for Modern Work

pAI (personal AI) is your own team of intelligent agents—custom-built to understand how you work and automate the tasks, decisions, and workflows that matter most to you.

mindmap
  root((pAI – Personal AI))
  
    Definition
      pAI is your personal team of intelligent agents that automate and augment your workflows.
    
    Core Principles
      Individual Context
      Composable
      Agentic
      Human-in-the-Loop
      Decentralized Ownership
      Interoperable

    Technology Stack
      Fast Agent
        Minimal Boilerplate
        Agent Framework
        Testing Support
      GitHub Actions
        Event-driven
        Scheduled Tasks
        Audit Trail
      Taskfile
        Development Tasks
        Operational Tasks
        Consistent Workflows
      MCP Protocol
        Gmail Integration
        Standardized APIs
        External Services
    
    Agent Organization
      @Home Context
        Gmail Curator
        Gmail Newsletter
        Test Agent
      @Work Context
        PR Diff Auditor
        Test Coverage Advisor

    Implementation Status
      Active Agents
        Gmail Curator
        Gmail Newsletter
      Experimental
        Test Agent
        New Capabilities
      Planned
        PR Diff Auditor
        Test Coverage Advisor
        DevOps Workflows
        SRE Automation

    Benefits
      Individual
        Reduced Email Overload
        Better Code Quality
        Automated Summaries
      Team
        Consistent PR Reviews
        Shared Visibility
        Knowledge Capture
      Org
        Scalable Automation
        Institutional Memory
        Quality Standards

    Next Steps
      Expand Agent Library
      Build Shared Utilities
      Implement Monitoring
      Create Documentation

    Challenges
      Context Management
      Token Costs
      Security
      Agent Coordination
      Feedback Loops

🧾 Definition

pAI (personal AI) refers to a system of intelligent agents custom-built to support an individual’s work by automating tasks, augmenting decision-making, and aligning with their personal workflows, values, and thinking patterns.

It's not "AI you use"—it's your AI team, working with and for you.

🧩 Core Principles

• Individual Context: Agents are shaped by your goals, standards, and preferences—not generic defaults.
• Composable: Built from lightweight workflows, prompts, and tools that can evolve.
• Agentic: Each agent acts semi-autonomously, handling tasks you delegate.
• Human-in-the-Loop: You remain the decision-maker; the AI operates as support staff, not a replacement.
• Decentralized Ownership: Individuals own the automations relevant to their role.
• Interoperable: Agents operate across systems (code, cloud, messaging, docs) with unified memory and action.

🧱 Platform: Fast Agent & GitHub Actions

• Fast Agent: Framework for defining, prompting, and testing AI agents with minimal boilerplate
• GitHub Actions: Event-driven runtime for scheduling and automation
• Taskfile: Modern task runner for development and operational workflows
• MCP (Model Context Protocol): Standardized way for agents to interact with external services
• Composable: Lightweight agents that can be combined and extended
• Auditable: Logs, permissions, and history make pAI systems inspectable and secure

🧠 Agent Organization

The repository is organized into two main contexts:

@Home - Personal Life Automation

• Gmail Curator: Monitors inbox for important non-newsletter emails and provides summaries
• Gmail Newsletter: Processes newsletters, extracts key articles, and delivers weekly digests
• Test Agent: Experimental agent for testing new capabilities

@Work - Professional Workflow Enhancement

• PR Diff Auditor: Analyzes pull request diffs for security issues, code quality, and best practices
• Test Coverage Advisor: Monitors test coverage trends and provides actionable testing recommendations

Each agent includes:

• Agentfile: Agent configuration and metadata
• Taskfile.yml: Development and operational tasks
• README.md: Agent-specific documentation
• agent/: Core agent implementation with FastAgent framework

🚀 Getting Started

🤖 Agent-Assisted Setup

You can use the GitHub Copilot agent or another agentic coder to set up RepoRadio CLI automatically.

Quickstart:

1. Download the setup instructions:

   curl -o PAI_INSTRUCTIONS.txt https://raw.githubusercontent.com/o3-cloud/pai/refs/heads/main/LLM.txt

2. Give the PAI_INSTRUCTIONS.txt file to your agent (e.g., Copilot agent) and prompt:

   Follow these instructions to setup pAI

The agent will guide you through the setup process interactively.

Prerequisites

• Taskfile - Modern task runner
• Agentman - Agent scaffolding tool
• OpenAI API key for LLM capabilities

Repository Structure

pAI/
├── @Home/              # Personal life automation agents
│   ├── gmail-curator/  # Email monitoring and summarization
│   ├── gmail-newsletter/ # Newsletter processing and digests
│   └── test-agent/     # Experimental agent testing
├── @Work/              # Professional workflow agents
│   ├── pr-diff-auditor/ # PR analysis and security scanning
│   └── test-coverage-advisor/ # Test coverage monitoring
└── docs/
    ├── README.md       # This file
    ├── STACK.md        # Technology stack details
    └── specs.md        # Specs system documentation

Running an Agent

Each agent can be run locally using Taskfile:

cd @Home/gmail-curator
task run

Or deployed to GitHub Actions for automated scheduling.

🧠 Mindset Shifts

🔄 Current Implementation Status

✅ Implemented Agents

🔄 Architecture Patterns

• Trigger Types: schedule (cron), workflow_dispatch (manual), pull_request (reactive)
• Autonomy Levels:
  • Informative: Email summaries, coverage reports
  • Suggestive: PR comments with recommendations
  • Autonomous: Scheduled processing and notifications
• Feedback Loops: Each agent can be tuned via ME.md context files and prompt engineering

✨ Benefits

For the Individual

• Reduces repetitive work
• Codifies personal expertise
• Enhances clarity, focus, and flow

For the Team

• Accelerates onboarding
• Improves shared visibility
• Unlocks reusable patterns

For the Org

• Multiplies impact of high performers
• Captures institutional knowledge
• Increases system resilience and speed

🛣 Next Steps

Current Focus

• Expand Gmail agents with more sophisticated filtering and categorization
• Implement PR Diff Auditor for GitHub pull request analysis
• Build Test Coverage Advisor with trend analysis and automated suggestions
• Create shared agent library for common patterns and utilities

Future Roadmap

• Add more @Work agents for DevOps and SRE workflows
• Implement cross-agent communication and shared context
• Build web dashboard for agent monitoring and configuration
• Create agent marketplace for sharing and discovering new agents

Contributing

• Each agent has its own README with setup and development instructions
• Use task commands for consistent development workflows
• Follow the FastAgent framework patterns for new agent development

⚠️ Current Challenges

📖 Documentation

• Specs System - Learn about the structured AI output framework that transforms unstructured content into actionable JSON data using 14 specialized extractors
• Technology Stack - Detailed information about the underlying technologies

🧩 Bonus: One-Line Summary

pAI is your personal team of intelligent agents—working behind the scenes to automate the tasks, decisions, and workflows that matter most to you.