ERP systems traditionally have been excellent at storing data and managing workflows, but not particularly adept at interpreting context or anticipating user needs. We’ve seen it many times: The system doesn’t always "understand" what it is holding.
With Odoo v18.3 (and v19), the introduction of AI-driven fields aims to bridge this gap, providing context-aware intelligence within the ERP. Instead of fields simply holding static data, these fields use AI models to analyze related information within the ERP automatically.
Let’s break down what’s new, how it works, and what it means.
The star of the show: AI fields
At the heart of this shift is a new concept: AI fields. These are regular Odoo fields—text, boolean, tags, many2one, etc.—that can be auto-populated by AI, based on the context you define.
The magic is in the prompt: you give the field a set of instructions (“summarize this help desk ticket in 2 sentences”) and tell it which fields in the record to reference (\description, \message_ids, etc.).
It’s like having GPT living inside your Odoo model schema.
You can use AI fields in:
Odoo Studio (drag-and-drop AI field into a form view)
Field properties (mark any field as AI-enabled)
Automations (compute a field with AI on trigger)
Under the hood: How it works
Creating an AI field involves three steps:
Define a prompt: What should the AI do?
Inject context: Reference other fields (\name, \email, \description, etc.)
Choose when to run it: Manually, on field change, or via automation.
From a tech perspective, Odoo connects to external LLMs (via OpenAI or Google API keys) and pipes data into them with your prompt. It handles the response and writes it back to the field.
Concrete examples with real outcomes
Let’s ground this with real use cases pulled straight from the 18.3 demo by Kevin Zaki:
🏷️ Use Case 1: Auto-tagging Leads by Industry
Instead of asking reps to manually tag opportunities by industry, AI analyzes the company name and email and assigns relevant tags like “Software”, “Retail”, or “Logistics”.
Prompt sample: “Classify the company into one of the following industries: Software, Retail, Manufacturing…”
🎯 Why it matters: Better segmentation, routing, and reporting—without humans in the loop.
😡 Use Case 2: Sentiment Detection on Tickets
Create a sentiment_score field that analyzes the tone of chatter messages on a helpdesk ticket.
Prompt: “Rate sentiment from 1 (very negative) to 5 (very positive) based on messages in \message_ids.”
🎯 Why it matters: Managers can auto-flag frustrated customers, trigger escalation, or prioritize replies.
🧑💼 Use Case 3: Smart Escalation Summaries
When a ticket is escalated, generate a 2-sentence executive summary using the original description and customer messages.
🎯 Why it matters: Avoid context loss. Save senior staff from reading 50+ back-and-forth messages.
💬 Use Case 4: Drafting Replies in CRM & Helpdesk
When a customer message comes in—via chatter, email, or contact form—AI can draft a suggested reply based on the full context of the conversation and related fields (e.g. lead stage, ticket type, recent orders).
Prompt sample: “Draft a professional reply to the latest customer message in \message_ids, using context from \lead_stage and \customer_order_history.”
🎯 Why it matters: Saves reps time, ensures tone consistency, and speeds up response times. Reps can edit instead of writing from scratch.
⚙️ Use Case 5: Smart Task Assignment by Skillset
Odoo can use AI to route tasks to the right team or user based on task description, context, and user roles. For instance, if a ticket mentions “invoice discrepancy,” the system assigns it to finance. If it includes “server outage,” it routes to IT support.
Prompt: “Based on the task description and tags, assign this task to the most relevant team.”
🎯 Why it matters: Reduces manual triage. Tasks get to the right person faster. Keeps specialists focused on what they’re best at.
🧠 Use Case 6: AI-Assisted Knowledge Retrieval in Helpdesk
When working on a ticket, support agents can click an AI button that queries the Knowledge app, extracts relevant documentation, and inserts links or answers directly into the ticket.
Prompt: “Search Knowledge Base articles for the top 3 related resources to help with this issue based on \description and \message_ids.”
🎯 Why it matters: Cuts time spent searching for docs. Accelerates support without switching context.
Why this is a big deal
AI fields shift ERP systems from systems of record to systems of action.
Instead of just storing information, Odoo now starts interpreting and acting on it.
This means:
Less manual data entry
Faster decision-making
More contextual automation
Reduced need for custom dev for edge cases
And it’s just the beginning. As LLMs get smarter and Odoo expands agent capabilities, we’ll see even more tasks get delegated to AI—especially in high-volume, low-context areas like support, marketing, and sales ops.
Bonus: AI Agents
Beyond AI fields, Odoo now also supports custom AI Agents—chatbots powered by LLMs like GPT-4 or Gemini that you can configure inside the system.
You can:
Set the agent’s tone (analytical, balanced, creative)
Attach documents (e.g. enterprise agreements)
Define specific tasks (e.g. schedule a meeting)
It’s early days, but the agent framework lays the foundation for internal assistants that know your company’s data and speak your company’s language.
If you’re exploring how to bring this level of intelligence into your workflows—whether it’s sales, support, ops, or beyond—bob can help. We’re a boutique Odoo partner obsessed with operational excellence, and we’ve already helped multiple clients implement AI-native ERP features that actually move the needle.
Book a meeting with us to explore what’s possible.
