🌶️ ChiliHead OpsManager v2.1 - OHHH1MAIL Platform

AI-powered restaurant operations management platform for Chili's #605 Auburn Hills

Complete email automation + AI agent orchestration + Visual debugging + Intelligent chat assistant

🎯 What Is This?

ChiliHead OpsManager v2.1 automates restaurant operations for John at Chili's #605 in Auburn Hills. The system:

Monitors Gmail for restaurant-related emails (filtered by domain/sender)
Processes emails through 5 specialized AI agents orchestrated by Apache Dolphin Scheduler
Extracts tasks and deadlines automatically from leadership communications
Provides AUBS - Your friendly AI assistant with complete operational awareness
Visualizes everything through Open-WebUI with embedded debugging tools

✨ Key Features

🤖 AUBS - Your Personal Assistant

AUBS (Auburn Hills Business System) is your friendly, intelligent operations assistant who:

Has complete awareness of all emails, tasks, and deadlines
Understands Chili's operations, RAP Mobile metrics, and HotSchedules
Prioritizes like you would (leadership deadlines first, then team issues, then guests)
Talks conversationally - not like a robot
Available 24/7 via chat in Open-WebUI

See full AUBS Guide

📧 Smart Email Processing

5 AI Agents working together:
- Triage Agent: Categorizes and routes emails
- Vision Agent: OCR for invoices, receipts, PDFs
- Deadline Scanner: Extracts dates and deadlines
- Task Categorizer: Identifies action items
- Context Agent: Synthesizes everything with 30-day memory
Email Filtering: Only processes emails from chilis.com, brinker.com, hotschedules.com
Attachment Handling: OCR, PDF parsing, image analysis

🎨 Open-WebUI Interface

8 tabs for complete operational visibility:

Dashboard - Overview of operations
Emails - Inbox, processed emails, agent outputs, quick actions
Tasks - Task management with priorities and due dates
Calendar - Events, deadlines, scheduling
Debug - UI-TARS execution timeline, agent decisions, screenshots, DAG visualization
Analytics - Performance metrics, agent stats, email volume, response times
Agents - Agent status, model selection, performance, logs
Settings - Model config, thresholds, notifications, integrations

🔄 Enterprise Orchestration

Apache DolphinScheduler 3.2.0 - DAG-based workflow orchestration
Fault-tolerant - Automatic retries and error handling
Visual debugging - UI-TARS embedded in Open-WebUI
GPU acceleration - NVIDIA GPU support for Vision Agent

🧠 Multi-Provider LLM Support

Easy model swapping between:

OpenAI: GPT-4o, GPT-5
Anthropic: Claude Sonnet 4 (AUBS uses this)
Ollama: Llama 3.2, local models

Configure per-agent in .env file.

🏗️ Architecture

┌─────────────────────────────────────────────────────────────┐
│                   Open-WebUI (Port 3040)                    │
│  Dashboard | Emails | Tasks | Calendar | Debug | Analytics  │
│                        Chat with AUBS                        │
└────────────────────────────┬────────────────────────────────┘
                             │
                             ▼
┌─────────────────────────────────────────────────────────────┐
│              AUBS (Autonomous Unified Business System)       │
│        Orchestrator | DAG Builder | Action Router           │
│               Chat Service (Claude Sonnet 4)                 │
└────────┬────────────────────────┬────────────────────────────┘
         │                        │
         ▼                        ▼
┌──────────────────┐    ┌──────────────────────────────────┐
│ Email Ingestion  │    │  Apache Dolphin Scheduler 3.2.0  │
│  - IMAP Monitor  │    │   - DAG Execution                │
│  - Domain Filter │    │   - GPU/CPU Workers              │
│  - Attachment DL │    │   - Task Scheduling              │
└──────────────────┘    └──────────────┬───────────────────┘
                                       │
         ┌─────────────────────────────┼────────────────────┐
         │                             │                    │
         ▼                             ▼                    ▼
┌─────────────────┐         ┌─────────────────┐  ┌─────────────────┐
│ Triage Agent    │         │  Vision Agent   │  │ Deadline Agent  │
│ (Llama 3.2)     │         │  (GPT-4o + GPU) │  │ (Llama 3.2)     │
└─────────────────┘         └─────────────────┘  └─────────────────┘
         │                             │                    │
         └─────────────────────────────┼────────────────────┘
                                       ▼
                            ┌─────────────────────┐
                            │  Task Agent         │
                            │  (Llama 3.2)        │
                            └──────────┬──────────┘
                                       ▼
                            ┌─────────────────────┐
                            │  Context Agent      │
                            │  (Claude Sonnet 4)  │
                            │  NO FALLBACK        │
                            └──────────┬──────────┘
                                       ▼
                ┌──────────────────────────────────────────┐
                │         MCP Tools (Actions)              │
                │  - Task Manager    - SMS Sender          │
                │  - Calendar Sync   - Email Responder     │
                └──────────────────────────────────────────┘

Infrastructure:
┌─────────────┬─────────────┬─────────────┬─────────────┬─────────────┐
│ PostgreSQL  │   Redis     │    NATS     │   Qdrant    │   Ollama    │
│  (Data)     │  (Cache)    │  (Events)   │ (Vectors)   │  (LLMs)     │
└─────────────┴─────────────┴─────────────┴─────────────┴─────────────┘

🚀 Quick Start (Windows)

Prerequisites

Docker Desktop with WSL2 enabled
NVIDIA GPU with CUDA support (for Vision Agent)
16GB+ RAM recommended
Gmail account with App Password configured
OpenAI API key (optional, for GPT models)
Anthropic API key (for Claude / AUBS chat)

Installation

Clone Repository

git clone https://github.com/johnohhh1/ohhhmail.git
cd ohhhmail

Configure Environment

Edit infrastructure/docker/.env with your credentials:
- OpenAI API key (if using GPT models)
- Anthropic API key (required for AUBS chat)
- Gmail credentials
- Email filtering (allowed domains/senders)
CRITICAL Email Filtering:
```
ALLOWED_DOMAINS=chilis.com,brinker.com,hotschedules.com
ALLOWED_SENDERS=
BLOCKED_DOMAINS=spam.com
BLOCKED_SENDERS=
```
Start All Services
```
start.bat
```
This will:
- Check Docker is running
- Start all 13 services via Docker Compose
- Wait for services to initialize
- Open Open-WebUI in your browser
Access Services
- Open-WebUI: http://localhost:3040
- Dolphin Scheduler: http://localhost:12345 (admin/admin)
- AUBS API: http://localhost:5000
- Grafana (optional): http://localhost:3001

💬 Chat with AUBS

Quick Start

Open http://localhost:3040
Click the chat button (floating button in bottom-right)
Start chatting!

Example Conversations

You: "What's urgent today?"

AUBS: "Hey John! Here's what needs your attention:

Leadership deadline - Your DM needs the manager schedule by Friday 5pm
Team issue - Sarah called out for tonight's shift
Guest complaint - Slow service feedback from last night

Want me to help you prioritize?"

You: "Help me plan my next 2 hours"

AUBS: "Sure! Based on what we have:

First 30 minutes:

Review and send manager schedule to DM (critical deadline)
Call Sarah to confirm her shift status

Next 60 minutes:

Address guest complaint - review section 3 service times
Prep for tonight with one server short

The DM deadline is most urgent. Want me to pull up the schedule?"

You: "What's our comp sales this week?"

AUBS: "I don't have access to RAP Mobile data yet (coming in next update), but I can see we processed 47 emails this week with 3 from corporate leadership. Want me to check if any mentioned sales numbers?"

See full AUBS Guide

📊 System Components

13 Docker Services

postgres - PostgreSQL 15 (persistent storage)
redis - Redis 7 (caching, state)
nats - NATS JetStream (event streaming)
qdrant - Vector database (context memory)
ollama - Local LLM hosting
dolphin-server - Dolphin scheduler server
dolphin-worker-gpu - GPU worker for Vision Agent
dolphin-worker-cpu - CPU workers for other agents
aubs - AUBS orchestrator and chat service
email-ingestion - Gmail monitoring and filtering
open-webui - Web interface (port 3040)
uitars - Visual debugging service
grafana (optional) - Monitoring dashboards

5 AI Agents

Agent	Model	Purpose	Timeout
Triage	llama3.2:8b-instruct	Email categorization	15s
Vision	gpt-4o (GPU)	OCR, image processing	30s
Deadline	llama3.2:8b-instruct	Date extraction	15s
Task	llama3.2:8b-instruct	Action identification	15s
Context	claude-sonnet-4	Context synthesis (NO FALLBACK)	30s

All agents configurable per environment in .env.

🛠️ Management Commands

Windows Batch Files

start.bat - Start all services
stop.bat - Stop all services
status.bat - Check service health
logs.bat - View service logs (interactive menu)
setup.bat - Initial setup and dependency installation

Docker Commands

# View all services
docker-compose ps

# View logs for specific service
docker-compose logs -f aubs
docker-compose logs -f email-ingestion

# Restart a service
docker-compose restart aubs

# Stop everything
docker-compose down

# Rebuild and restart
docker-compose up -d --build

⚙️ Configuration

Agent Models

Edit infrastructure/docker/.env to change models:

# Triage Agent - Fast categorization
TRIAGE_AGENT_PROVIDER=ollama
TRIAGE_AGENT_MODEL=llama3.2:8b-instruct
TRIAGE_AGENT_FALLBACK=openai:gpt-4o-mini

# Vision Agent - OCR (needs GPU)
VISION_AGENT_PROVIDER=openai
VISION_AGENT_MODEL=gpt-4o
VISION_AGENT_GPU=true

# Context Agent - CRITICAL, NO FALLBACK
CONTEXT_AGENT_PROVIDER=anthropic
CONTEXT_AGENT_MODEL=claude-sonnet-4
CONTEXT_AGENT_FALLBACK=DISABLED

Email Filtering (CRITICAL)

# Only process emails from these domains
ALLOWED_DOMAINS=chilis.com,brinker.com,hotschedules.com

# Optionally restrict to specific senders
ALLOWED_SENDERS=dm@chilis.com,regional@brinker.com

# Block specific domains or senders
BLOCKED_DOMAINS=spam.com,marketing.com
BLOCKED_SENDERS=noreply@spam.com

After changing filters, restart email-ingestion:

docker-compose restart email-ingestion

📁 Project Structure

ohhhmail/
├── aubs/                        # AUBS orchestrator
│   └── src/
│       ├── main.py             # FastAPI app
│       ├── orchestrator.py     # DAG builder
│       ├── dag_builder.py      # DAG construction
│       ├── action_router.py    # MCP tool routing
│       └── chat.py             # AUBS chat service ⭐
│
├── agents/                      # 5 AI agents
│   ├── triage/
│   ├── vision/
│   ├── deadline/
│   ├── task/
│   └── context/
│
├── email_ingestion/             # Gmail monitoring
│   ├── monitor.py              # IMAP monitoring
│   ├── processor.py            # Email processing ⭐ (filtering)
│   └── config.py               # Configuration
│
├── openwebui/                   # Open-WebUI components
│   ├── components/
│   │   ├── Dashboard.tsx
│   │   ├── EmailDashboard.tsx
│   │   ├── TaskManager.tsx
│   │   ├── CalendarView.tsx
│   │   ├── UITARSDebugPanel.tsx  # Embedded debugging
│   │   ├── Analytics.tsx
│   │   ├── AgentMonitor.tsx
│   │   ├── Settings.tsx
│   │   └── ChatWithAUBS.tsx    # AUBS chat UI ⭐
│   └── config/
│       └── tabs-config.json    # 8 tabs configuration
│
├── mcp_tools/                   # MCP action tools
│   ├── task_manager.py
│   ├── calendar_integration.py
│   └── sms_sender.py
│
├── shared/                      # Shared libraries
│   ├── models.py               # Pydantic models
│   ├── llm_config.py           # Multi-provider LLM
│   └── clients/                # Service clients
│
├── infrastructure/
│   ├── docker/
│   │   ├── docker-compose.yml  # 13 services
│   │   └── .env                # Configuration ⭐
│   └── dolphin/                # Dolphin configs
│
├── docs/                        # Documentation
│   ├── AUBS_GUIDE.md           # AUBS user guide ⭐
│   ├── BUILD_STATUS.md         # Implementation status
│   ├── IMPLEMENTATION_PLAN.md  # 10-week plan
│   └── QUICK_START.md          # Getting started
│
├── start.bat                    # Windows start script
├── stop.bat                     # Windows stop script
├── status.bat                   # Health check script
└── logs.bat                     # Log viewer script

⭐ = Recently updated with AUBS persona and email filtering

📚 Documentation

AUBS Guide - Complete guide to using AUBS
Quick Start - Getting started guide
Build Status - Implementation status
Implementation Plan - 10-week roadmap
Build Orchestrator - AI swarm build strategy

🔍 Monitoring & Debugging

Health Checks

# Check all services
status.bat

# Individual health endpoints
curl http://localhost:5000/health      # AUBS
curl http://localhost:8001/health      # Email Ingestion
curl http://localhost:12345/dolphinscheduler/health  # Dolphin

Logs

# Interactive log viewer
logs.bat

# Or direct Docker commands
docker-compose logs -f aubs
docker-compose logs -f email-ingestion
docker-compose logs -f dolphin-server

UI-TARS Debug Panel

Open http://localhost:3040
Click "Debug" tab
View:
- Execution timeline with timestamps
- Agent decisions and reasoning
- Screenshots from Vision Agent
- DAG visualization
- Replay capability

🎯 Current Status

✅ Completed

🚧 In Progress

PostgreSQL persistence for chat sessions
RAP Mobile metrics integration
HotSchedules integration
Google Calendar sync
ChiliHead 5-Pillar Delegations (can wait)

📋 Planned

🔒 Security & Privacy

Data Privacy

All data stays on your infrastructure
No external data sharing
Gmail credentials stored securely in environment variables
Email filtering prevents spam and noise

API Keys

OpenAI API key: Optional (only if using GPT models)
Anthropic API key: Required (for AUBS chat and Context Agent)
All API calls encrypted and ephemeral

Access Control

Currently single-user (John)
Can be extended with authentication

🐛 Troubleshooting

AUBS Not Responding

Check AUBS service:

docker-compose ps aubs
docker-compose logs -f aubs

Verify Anthropic API key in .env:
```
ANTHROPIC_API_KEY=sk-ant-api03-...
```
Restart AUBS:
```
docker-compose restart aubs
```

Emails Not Being Processed

Check email filtering configuration in .env
Verify Gmail credentials
Check email-ingestion logs:
```
docker-compose logs -f email-ingestion
```

Services Won't Start

Verify Docker Desktop is running
Check port 3040 is available (not 3000)
Review .env for missing values
Check logs: docker-compose logs

GPU Not Detected (Vision Agent)

Verify NVIDIA drivers installed
Check Docker Desktop GPU support enabled
Test: docker run --gpus all nvidia/cuda:11.0-base nvidia-smi

👨‍💼 About

Built specifically for:

John Olenski - Managing Partner
Chili's #605 - Auburn Hills, Michigan
Brinker International - Franchise operations

Contact:

Email: john.olenski@gmail.com
GitHub: @johnohhh1

📄 License

Built with ❤️ by a GM who believes great people deserve great tools.

🌶️ "Excellence Through Leadership & Accountability"

Name		Name	Last commit message	Last commit date
.claude		.claude
.github/workflows		.github/workflows
agents		agents
aubs		aubs
chilihead-app		chilihead-app
docs		docs
dolphin		dolphin
email_ingestion		email_ingestion
infrastructure		infrastructure
mcp-tools		mcp-tools
openwebui		openwebui
shared		shared
tests		tests
.gitignore		.gitignore
CURRENT_STATE_ASSESSMENT.md		CURRENT_STATE_ASSESSMENT.md
NEXT_STEPS.md		NEXT_STEPS.md
PROGRESS_NOTES.md		PROGRESS_NOTES.md
README.md		README.md
REBUILD_PLAN.md		REBUILD_PLAN.md
SHARED_UTILS_SUMMARY.md		SHARED_UTILS_SUMMARY.md
V2_MIGRATION_PLAN.md		V2_MIGRATION_PLAN.md
VISION.md		VISION.md
draftprd.md		draftprd.md
logs.bat		logs.bat
previousconvocontex.txt		previousconvocontex.txt
requirements-dev.txt		requirements-dev.txt
requirements.txt		requirements.txt
setup.bat		setup.bat
start.bat		start.bat
status.bat		status.bat
stop.bat		stop.bat

Latest commit

History

Folders and files

Repository files navigation