Odoo AI APIs: How to Connect External Models

Your team already lives in Odoo for CRM, projects, and operations. The gap is rarely "more software." It is faster answers, cleaner follow-ups, and Odoo automation that scales without adding headcount for every repetitive task.

Odoo AI is built into Odoo 19 to give context-aware assistance inside the same interface. When you need a specific provider or a custom chain, you connect AI in Odoo through documented settings and, where needed, integration work.

This guide explains what is native in Odoo (from the official Odoo AI documentation), how Odoo AI tools map to real workflows, and how an Odoo ChatGPT integration or Gemini setup fits next to agents and server actions.

For related reading on the same topic line, see our posts on Odoo AI and ChatGPT workflow automation and Odoo AI agents and business automation.

What is Odoo AI and connecting external models?

Quick answer: In Odoo 19, "connecting external models" for many teams means configuring the AI app to use provider APIs (documented for OpenAI ChatGPT and Google Gemini), picking an LLM when you build agents, and using integrations only when you need behavior outside those native options.

Odoo is your ERP and operations hub. External models are the large language models hosted by vendors such as OpenAI or Google, reached through API keys and the choices Odoo exposes in settings and in each agent.

Native path: Install the AI app where needed, configure providers under AI app, Configuration, Settings, and assign models to agents as described in the documentation. That is how Odoo AI ties to ChatGPT or Gemini without custom code.

Integration path: If you need another vendor (for example Anthropic Claude), a proprietary API, or orchestration Odoo does not ship, that is custom integration: your team or a partner wires HTTP calls, security, and monitoring around Odoo data and workflows.

How Odoo AI works in Odoo 19

Odoo documents AI as productivity assistance across apps: intelligent, context-aware help so users stay inside Odoo.

Ask AI (assistant)

• Press Ctrl+K to open the command palette, enter a prompt, then use the AI option to talk to the Ask AI agent.
• Use the AI button in the top-right; suggested prompts can change based on where you are in the database.
• Common requests include translating chatter, summarizing threads, generating follow-ups, improving drafts, and suggesting next steps for sales or support.
• After a reply, you can send to email, log as a chatter note, or copy to clipboard. Default prompts can be edited in the AI application.

The standard Ask AI agent does not modify the database: it can open views and help with content, not create leads or change records on its own. Custom agents with topics and tools are documented separately.

Providers and API keys (native configuration)

According to the AI API keys documentation, Odoo supports Gemini and OpenAI (ChatGPT) as providers in the AI application settings. You manage credentials and defaults there.

• On Odoo.sh or on-premise databases, API keys are required to use AI features.
• On Odoo Online, adding your own keys is optional; some organizations still add keys for policy or control.
• Using provider APIs may incur fees from the vendor; pricing depends on model and account.

Agents, automation, and workflows

• AI agents combine topics, tools, and sources. When you create an agent, you select an LLM model from a list; Odoo documents support for multiple versions of ChatGPT and Gemini.
• AI server actions let AI choose among tools (other server actions marked "Use in AI") during a workflow, with tools holding the Python that updates records.
• Other documented areas include AI in email templates, AI fields, live chat, voice transcription, document sort, support operations, and improving text. See the main AI index for the full list.

Together, these are the main levers for Odoo AI tools in the product: in-app assistance, provider-backed agents, and governed automation.

Key benefits for businesses

• Time savings: Less manual drafting in CRM, helpdesk, and email. Ask AI and template-based generation scale wording without leaving Odoo.
• Cost reduction: Fewer shadow tools and copy-paste errors when teams work from one system with clear AI entry points.
• Better decisions: Summaries and suggested next steps help managers focus on exceptions, not every thread from zero.
• Scalability: Agents and AI server actions support repeatable processes while keeping execution inside Odoo patterns you can audit.

Real Odoo AI use cases

Below are concrete examples. Native behavior follows Odoo 19 documentation; anything beyond listed providers is integration work.

1. Faster email and chatter (native)

Use Ask AI to improve drafts, summarize threads, or suggest follow-ups. Use AI in email templates where prompts are evaluated per record at send time.

2. Sales assistant (native UX, guarded data changes)

Reps get next-step suggestions and message help from Ask AI. Creating or updating CRM records uses agents with the right topics and tools, not the default Ask AI agent alone.

3. Accounting and document-centric flows (native patterns)

AI server actions can route work to tools that move or tag documents when your tools implement those rules in code.

4. Data enrichment (usually integration)

Third-party data APIs (firmographics, risk, enrichment) are typically custom integrations. Odoo AI does not replace licensed data providers by itself.

5. Support and live chat (native + configuration)

Odoo documents AI for support workflows and live chat; expect configuration and clear guardrails.

6. Choosing ChatGPT or Gemini per agent (native)

For multi-scenario rollouts, you assign an LLM per agent in the AI app. That is the supported way to align model choice with use case inside Odoo.

For CRM-focused ideas, you can also read Odoo AI and GPT-4 for CRM and sales. For website and data structures, our blog.post model guide explains how content is stored in Odoo.

Native Odoo AI vs external AI (ChatGPT, Claude)

Native (documented in Odoo 19): Ask AI, provider settings for OpenAI and Gemini, agents with LLM selection, AI server actions and tools, plus features linked from the main AI page (email templates, fields, live chat, voice, document sort, support, improve text).

Pros: One product surface, documented configuration, less glue for standard scenarios.

Cons: You work within what Odoo ships and documents. Exotic chains may need design beyond defaults.

External integrations: ChatGPT and Gemini are first-class in AI settings. Claude or other APIs are not described as built-in provider toggles in the same documentation; connecting them means custom modules or middleware, with your team owning keys, data flow, and monitoring.

Pros of custom integrations: Model choice and orchestration outside the native provider list.

Cons: Higher maintenance, security review, and cost tracking on your side.

Limitations and considerations

• Data quality: AI output reflects your masters, stages, and chatter discipline. Clean data beats bigger models.
• Implementation complexity: Agents need clear topics and tools. AI server actions need well-defined tools and prompts. Templates need legal and brand review.
• Costs: Plan for Odoo licensing, optional provider API fees, and partner time for integration.
• Security: Decide what may leave your boundary to a provider. Document roles, retention, and audit expectations. On-premise and Odoo.sh must supply keys per Odoo docs.

How to implement AI in Odoo

1. Audit: Map where time is lost and where errors repeat. Confirm Odoo apps in scope.
2. Identify use cases: Pick a small set with measurable outcomes. Prefer native Ask AI, templates, and agents before custom APIs.
3. Choose tools: Install the AI app if you need custom keys or agent authoring. Configure providers per the API keys documentation.
4. Integrate and test: Pilot with one team. Validate customer-facing text and financial outputs.
5. Optimize: Refine prompts, tools, and training. Scale what works.

Experts shorten this cycle: fewer dead ends, clearer acceptance tests, safer rollouts.

How we help companies implement Odoo and AI

Dasolo implements Odoo, connects systems, and automates operations. For AI in Odoo, we align native features with your processes and design integrations when external models or APIs are the right fit.

• Implementation: Solid foundations, clean configuration, workflows users adopt.
• Integrations: Reliable links between Odoo and your stack, including provider-backed AI where documented.
• Automation: Server actions, workflows, and AI-assisted patterns grounded in your data model.
• Optimization: Measurement, iteration, and governance as you grow.

We keep recommendations practical: what Odoo documents today, what is customization, and what is integration.

Conclusion

Odoo AI gives SMEs a direct path to assist users inside Odoo, configure ChatGPT and Gemini through documented provider settings, and build governed automation with agents and AI server actions.

The next step for many teams is not chasing every new model. It is clear processes, reliable data, and measured rollouts. ERP and AI together work best when workflows are owned and improved over time.