DigitalFTE is a local-first AI operations system for inbox triage, approvals, social drafting, finance workflows, and weekly briefings.

🏆 Built at a 48-hour hackathon — what started as a weekend project to automate my overflowing inbox turned into a full autonomous agent system.

This repo now includes a real local control center:

• a polished browser UI for queue review, approvals, setup readiness, and recent activity
• one-click generation of briefings and vault dashboard refreshes
• quick capture so you can create new work items without hand-editing markdown
• repaired startup scripts and compatibility wrappers so the repo launches coherently

• 📧 Email - Monitors Gmail, drafts replies in your voice, you approve before sending
• 💬 WhatsApp - Receives messages, generates contextual responses
• 📱 Social Media - Auto-posts to LinkedIn, Twitter, Facebook, Instagram
• 💰 Accounting - Creates invoices & bills in Odoo, generates P&L reports
• 📊 Weekly Briefing - Automated summary of revenue, tasks, and metrics

This is the magic. DigitalFTE doesn't send generic AI-sounding messages — it learns YOUR voice.

How it works:

1. Create /vault/EmailStyle.md with your writing personality:

# My Writing Style

## Tone
Casual but professional. I use "Hey" not "Dear". Short sentences.

## Phrases I Use
- "Let me know if that works"
- "Happy to jump on a call"
- "Cheers" (not "Best regards")

## Things I Never Say
- Corporate jargon
- "I hope this email finds you well"

2. The AI reads this before drafting ANY message — emails, WhatsApp, tweets, LinkedIn posts
3. Every response sounds like you actually wrote it

Works across all channels:

• 📧 Email replies match your professional tone
• 💬 WhatsApp messages match your casual texting style
• 🐦 Tweets capture your social media voice
• 💼 LinkedIn posts maintain your industry presence

You can create separate style guides for each channel, or use one unified voice.

# install python deps
python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

# configure credentials as needed
cp .env.example .env

# start the local stack
./start_all.sh

Then open:

• Control Center: http://127.0.0.1:8282
• Odoo UI: http://localhost:8069 if Docker is available
• Vault dashboard: vault/Dashboard.md

Stop everything with:

./stop_all.sh

The control center is the main operator surface. It gives you:

• live queue views for Needs_Action, Pending_Approval, Approved, Rejected, and Done
• full preview of vault items and one-click moves between stages
• setup visibility for Gmail, WhatsApp, social, finance, and OpenAI
• quick capture for new tasks and manual requests
• briefing generation and dashboard refresh from the UI

Email arrives → Gmail Watcher detects it → AI drafts reply (in your style)
    → You review in vault/Pending_Approval/ → Approve → Sent

The same vault-native flow works for WhatsApp, social media, and manual operator tasks.

┌─────────────────────────────────────┐
│  Watchers (Gmail, WhatsApp, etc.)   │
└──────────────┬──────────────────────┘
               ↓
┌─────────────────────────────────────┐
│  Obsidian Vault (your local brain)  │
│  - Style guides                     │
│  - Pending approvals                │
│  - Action history                   │
└──────────────┬──────────────────────┘
               ↓
┌─────────────────────────────────────┐
│  Orchestrator + AI                  │
│  - Reads your style                 │
│  - Drafts responses                 │
│  - Routes for approval              │
└──────────────┬──────────────────────┘
               ↓
┌─────────────────────────────────────┐
│  MCP Servers (Gmail, Twitter, etc.) │
│  - Executes approved actions        │
└─────────────────────────────────────┘

• Python 3.13+
• Docker Desktop for Odoo/Postgres
• API keys as needed: OpenAI, Gmail OAuth, Twilio, LinkedIn, Twitter/X, Meta

DigitalFTE/
├── control_center/        # Local FastAPI app + polished operator UI
├── vault/                 # Obsidian-native working memory and approvals
├── agents/                # Watchers, orchestrators, sync, webhook server
├── watchers/              # Compatibility imports for legacy paths
├── scripts/               # Launch entrypoints
├── utils/                 # AI drafters and helpers
└── mcp_servers/           # API integrations

# start only the control center
python3 scripts/control_center.py

# generate a new weekly briefing
python3 scripts/weekly_audit.py

# run tests
pytest -q

MIT