AI Vision Quality Control
Catch visual defects on the shop floor automatically — powered by Claude, GPT‑5.6, or Groq, right inside Odoo's native Quality app.
4
AI providers supported100%
Human confirms every result1-click
Analyze from any tabletFull
Audit trail per checkWhy AI Vision Quality Control?
🔍 Native to Quality
No separate app to learn. Adds an AI tab straight onto your existing Quality Points and Quality Checks — the same screens your operators already use every day.
🤖 Bring your own AI
Connect Anthropic Claude, OpenAI, Groq, or a self-hosted OpenAI-compatible vision endpoint. Mix a fast/cheap primary provider with a stable fallback for production lines.
✅ Human always decides
The AI never silently sets Pass/Fail. It proposes a verdict, confidence score, and itemized defects — a person always confirms the final call.
See it in action
A full walkthrough: configuring a provider, enabling AI on a Quality Point, capturing a photo, and reviewing the verdict.
How it works
1. Configure a provider
Add Claude, OpenAI, Groq, or a self-hosted endpoint. Set the model, API key, and image/token limits once — reuse it across every Quality Point.
2. Enable AI on a Quality Point
Upload reference (golden-sample) images, pick the defect categories to watch for, and set a confidence threshold for routing borderline results to a human.
3. Capture & analyze
On the shop floor, snap a photo on the Quality Check and tap Analyze with AI. Works from a tablet, same as any native Odoo Quality step.
4. Review & confirm
Get a Pass / Fail / Needs Review verdict with a confidence score and itemized defects. Fails automatically raise a Quality Alert — the operator confirms the final result.
Built for production reliability
🔁 Automatic fallback
Configure a second AI provider per Quality Point. If the primary call fails, the check automatically retries on the fallback before ever reaching "Needs Review" — and a manual "Retry with Fallback" button is always available for a second opinion.
📊 Full audit log
Every AI call — provider, model, tokens, latency, verdict, raw response — is logged per Quality Check. Export it for compliance reviews in regulated industries.
🔒 Data control
Point at a self-hosted, OpenAI-compatible vision endpoint (e.g. Ollama) to keep factory-floor photos entirely on your own infrastructure.
⚙ Cost & latency controls
Per-provider image resizing and token limits keep every analysis fast and cheap — tune them per rate-limit tier without touching code.
Get started in minutes
Install the module, add an AI provider under Quality > Configuration > AI Vision Providers, enable AI on any Quality Point, and you're ready to catch defects before they ship.
| Availability |
Odoo Online
Odoo.sh
On Premise
|
| Odoo Apps Dependencies |
•
Discuss (mail)
• Quality (quality_control) • Inventory (stock) |
| Lines of code | 871 |
| Technical Name |
quality_ai_vision |
| License | LGPL-3 |
| Website | https://www.linkedin.com/in/saidyahia |
Please log in to comment on this module