# Odoo Workflow Platform

AI-assisted workflow, scheduling and resource management platform for two initial customers. Built on Odoo Community Edition with AI agent layer via n8n and local Ollama inference.

## Customers

### Customer 1 — Marquee Company Erection and event infrastructure company.

Requirements: - Crew scheduling — which teams go to which events - Resource/stock planning — tents, poles, weights, sidewalls, flooring - Project timeline management — setup, event, teardown days - Customer quotes to job to resource allocation to crew scheduling pipeline - Teams integration for crew communications - 365 calendar for job dates and planning - Weather awareness (future agent feature)

### Customer 2 — L8 Water Testing Company Legionella/L8 compliance testing and certification.

Requirements: - Scheduled testing visits (legally mandated intervals) - Engineer routing and scheduling - Certificate generation and tracking - Compliance deadline alerts — missing L8 test dates is a legal liability - Asset register — sites, water systems, last tested, next due - 365 calendar for engineer schedules - Automated compliance reports to clients via email

—

## Architecture

### Hosting Split

Hetzner VPS handles production Odoo — two paying customers cannot depend on local hardware for business-critical scheduling. AI/agent layer runs locally on kaburuaibox or Z600, calls Odoo via API.

### Odoo Multi-Tenant Both customers run as separate Companies within one Odoo instance. Data fully isolated by company. One codebase, one maintenance overhead.

—

## Odoo Modules Required

—

## Integrations

### Microsoft 365 / Teams - Microsoft Graph API for calendar, email, Teams messages - Azure app registration required per customer (or one multi-tenant app) - Odoo native Microsoft 365 sync module (Community) - Teams webhooks for job assignment notifications

### Email - Brevo preferred for outbound transactional email — delivery tracking, templates, compliance - 365 for direct customer-facing business email where needed

—

## AI Agent Layer

Agents built in n8n, calling Odoo API, Microsoft Graph API, Ollama, and Brevo API.

### Example Agent Workflows

Marquee: - Who is available for a job on a given date — check calendar, suggest crew, notify Teams - Resource conflict detection — same equipment double-booked - Weather check on job week — alert crew lead if bad forecast - Quote to job creation automation

L8 Water Testing: - Sites due testing in next 30 days — list and nearest engineer - Compliance deadline alerts — auto-email customer 2 weeks before due date - Certificate generation on job completion - Missed test date escalation alerts

—

## Phased Delivery

### Phase 1 — Core Platform - [ ] Odoo Community on Hetzner VPS - [ ] Configure both companies (multi-tenant) - [ ] Field Service, Project, Calendar modules - [ ] 365 calendar sync - [ ] Brevo email integration - [ ] Basic job/scheduling workflows - [ ] User training

### Phase 2 — Agent Workflows - [ ] n8n on kaburuaibox or Z600 - [ ] Odoo API connector in n8n - [ ] Microsoft Graph API integration (Azure app registration) - [ ] Teams notification workflows - [ ] Scheduling reminder agents - [ ] L8 compliance deadline alerts

### Phase 3 — AI Layer - [ ] Natural language queries against Odoo data - [ ] Auto-draft customer emails - [ ] Resource conflict detection - [ ] Intelligent job scheduling suggestions - [ ] Expansion per customer needs

—

## Infrastructure Notes

### Z600 Role HP Z600 (dual Xeon E5606, 24GB DDR3, MSI RTX 3060 12GB) earmarked as dedicated node for AI/agent layer. Runs Ollama + n8n, calls Hetzner-hosted Odoo via API. Does NOT host Odoo itself.

### Hetzner VPS Sizing (TBD) Odoo Community for small deployment needs approximately 4 vCPU, 8GB RAM, 80GB SSD. CCX23 or similar. Separate PostgreSQL or co-hosted initially.

—

## Commercial Notes - Both customers expandable beyond initial scope - Phases can be sold and delivered independently - L8 compliance module may have wider market value as standalone product - Consider white-labelling per customer