Free Odoo MCP: connect Odoo to ChatGPT, Claude, Gemini and MCP-compatible AI agents.

What if you could ask your AI assistant:

• “Show me our top 20 customers by revenue this year.”
• “Which sales orders are still waiting to be invoiced?”
• “Find all products with low stock and summarise what needs attention.”

Instead of copying data from Odoo into ChatGPT, Claude or Gemini, an Odoo MCP server gives the AI a structured way to work with your Odoo environment.

That's what Free Odoo MCP is for. It connects Odoo with MCP-compatible clients — ChatGPT, Claude, Gemini, Cursor and other agents — so those tools can discover Odoo capabilities and use them when you ask.

What is Odoo MCP?

Odoo MCP means connecting Odoo to the Model Context Protocol (MCP). MCP is an open protocol that lets AI applications talk to external tools, services and data sources through a common interface — instead of a custom integration for every assistant. The longer definition is what is Odoo MCP?; the working guide is how to use MCP with Odoo.

In simple terms:

Odoo → MCP → AI assistant or AI agent

The assistant can then discover the available Odoo tools and use them when answering a request: search, read, and — if you allow it — create, update or delete records. Several open-source MCP servers already expose that surface over Odoo's APIs; ours runs inside Odoo so it does not need the Custom-plan external API. See what Odoo API access actually costs.

Why connect Odoo to AI?

Odoo already holds the business: customers, vendors, sales orders, invoices, products, inventory, CRM, purchases, employees, projects, accounting, manufacturing, helpdesk, website and eCommerce.

Getting a useful answer out of that usually means menus, filters, exports, SQL or a custom report. An AI agent changes the interaction. Instead of “where is that report?”, you ask “which customers generated the most revenue last quarter?” Instead of searching unpaid invoices by hand, you ask for invoices older than 60 days and get a summary.

That's why MCP interest is growing among Odoo developers, implementers and operators — reporting, automation, debugging and natural-language access without a new integration for every tool.

Free Odoo MCP works with your AI stack

You are not locked to one vendor. The same Odoo MCP layer can serve ChatGPT, Claude, Gemini, Cursor and other MCP-compatible clients — if that client actually supports the transport.

ChatGPT

Use ChatGPT as a natural-language interface to Odoo — plan limits are in OpenAI’s MCP apps docs. Example: “Summarise this month's sales pipeline and identify opportunities that haven't been contacted in the last 14 days.” How to connect Odoo to ChatGPT — and note that not every ChatGPT plan supports every MCP feature.

Claude

Claude is the client most technical Odoo users start with. Anthropic’s remote MCP connectors are the product surface. Give it structured MCP access instead of one-off exports and scripts. How to connect Odoo to Claude · Can Claude access Odoo?

Watch on YouTube · Open the live sandbox

Gemini

MCP-compatible Gemini tools and agents can use the same Odoo layer — Google documents remote MCP and allowed_tools — so you can change models without rebuilding the integration. How to connect Odoo to Gemini.

MCP-compatible AI agents

That's the point of MCP. You are not building “an Odoo chatbot for one vendor.” You are building an AI-access layer for Odoo — coding agents, internal assistants, Cursor and other MCP-compatible clients. How to connect an AI agent to Odoo · Can AI agents access Odoo?

What can an AI agent do with Odoo?

Exact tools depend on the MCP server and the permissions you grant. Common work:

1. Ask questions about the business

• “How much revenue did we generate this month?”
• “Who are our five largest customers?”
• “Which quotations haven't been followed up?”
• “How many products are below reorder level?”
• “Show open CRM opportunities over $10,000.”

The model turns that into structured queries against Odoo.

2. Analyse Odoo data

Not just a record dump. “Analyse our pipeline and tell me where we lose opportunities.” “Compare this month's sales with last month and explain the biggest changes.” That is more useful than a traditional chatbot glued to one screen.

3. Generate reports

Management summaries, sales and inventory snapshots, customer analysis, receivables, CRM, exception lists — asked in English rather than learned as Odoo report screens.

4. Find problems

Exception workflows: overdue invoices with open opportunities; products with declining sales and rising stock; orders with unusual delays. The agent can join models and explain what it found.

5. Automate repetitive work

With tightly scoped permissions, some MCP servers can create or update records, not only read them. Write access needs more care than read access. Start with AI proposes → human reviews → Odoo changes, not AI decides → Odoo changes immediately.

Is Odoo MCP safe?

MCP is not a permission system. Your Odoo rights still apply. The server acts as an authenticated Odoo user or API credential — treat that like any other powerful integration login.

• Start read-only. Read → analyse → report, before create → update → delete.
• Use a dedicated user. Not the administrator. Only the models the workflow needs. Several MCP implementations say the same: realistic Odoo groups, not admin.
• Test on staging. Connect a copy, try good prompts, bad prompts and edge cases, review the actions, then consider production.
• Keep writes gated. Approval, audit logs, restricted models, short-lived tokens.

Our product page is the governance detail: scopes, rate limits, write approval and an audit row per call. Watch the controls walkthrough or see the AI-agents industry setup.

Odoo MCP vs a traditional Odoo API integration

XML-RPC, JSON-RPC and Odoo 19's JSON-2 API are still the right tools for scheduled syncs and batch work. MCP does not replace them. It is an AI-facing layer on top. New MCP work on Odoo 19 should target JSON-2, not the deprecated RPCs.

A conventional integration says “call this endpoint with these parameters.” An MCP client can see the available tools and pick the one that matches the request. That is why it fits assistants, coding agents, natural-language reporting and internal copilots.

Full cost of the API gate: Odoo external API access.

Why use a free Odoo MCP?

You should not need another SaaS bill just to try AI against Odoo. Start small: connect an instance, give the agent limited rights, ask questions, then decide what becomes a production workflow.

You can try our server live on sample data with no signup. On an Odin deployment the module is included in the hosting fee. If you self-host, it is also on the Odoo Apps Store.

Community and Enterprise

Open MCP projects target both editions. Check the specific server against your Odoo version, Odoo Online vs Odoo.sh vs on-premise, API availability and authentication. Do not assume every server supports every deployment.

Odoo Online, Odoo.sh, or self-hosted?

Some MCP servers run outside Odoo and call the API. Others are Odoo modules. An external server may work without installing a module; a native module can go deeper and, in our case, avoid the Custom-plan API gate. Before you install anything, confirm version, hosting, whether the API is exposed, whether you need a public HTTPS MCP endpoint, and how auth works. Odoo MCP for Odoo Online · Is there a free Odoo MCP server?

Why MCP is interesting for Odoo developers

The old pattern is one integration per use case: a report feed, then a chatbot, then another assistant. MCP lets you expose one set of Odoo capabilities to every compatible client. Same underlying tools, different AI front-ends. That is why the ecosystem is moving quickly — different architectures, auth models and Odoo versions, same idea.

Get started with Free Odoo MCP

Your customers, sales, stock and invoices already live in Odoo. MCP is how the AI tools you already use can work with that data.

Start with a read-only connection, test it, then decide what should be allowed to write.

Try it live — no signup · Watch the Claude demo · Odoo MCP product · Book a call

Official documentation

• Model Context Protocol
• Claude: custom connectors with remote MCP
• OpenAI Help: developer mode and MCP apps in ChatGPT
• Gemini API: function calling and remote MCP
• Odoo 19 External API (JSON-2)
• Odoo release notes (19.4, July 2026)

Keep going