From Answering Questions to Doing the Work

Artificial intelligence has already changed how businesses search for information, create content, and analyse data.

But there is a difference between an AI tool that answers a question and an AI system that can take action inside your business processes.

That distinction is central to the AI story in Odoo 20.

With the release of Odoo 20, AI is moving deeper into the ERP itself. Odoo's AI agents can understand natural-language instructions, work with information from the Odoo database, use configured tools, create and update records, and participate in automated workflows.

Odoo's official release announcement describes the new capability in practical terms: businesses can instruct an AI agent to automate a process, have it explain what it is about to do, and then put that process into operation.

Examples include automatically confirming a sales order when its matching email arrives, assigning leads to the appropriate salesperson, creating a project when a contract is signed, or notifying a team when stock reaches a defined level.

This changes the conversation around AI in ERP.

The question is no longer simply:

"Can AI help my employees work faster?"

It becomes:

"Which business processes can AI help us run?"

For organisations considering Odoo 20, understanding this distinction is important because AI agents can affect sales, finance, inventory, purchasing, projects, documents, customer service, websites, and other areas of the business.

What Is an AI Agent in Odoo 20?

An AI agent in Odoo 20 is a virtual assistant that can understand natural language and interact with Odoo tools to perform tasks.

Think of it as a digital worker operating within defined boundaries.

A traditional chatbot might answer:

"What are our outstanding invoices?"

An AI agent can potentially go further.

It can:

1. Understand the request.
2. Access the relevant Odoo information.
3. Use an appropriate tool.
4. Retrieve or modify records where permitted.
5. Return the result to the user.

The important part is action.

Odoo's documentation explains that agents are built around Skills and Sources.

Skills

Skills define what the agent is capable of doing. They contain instructions and the tools the agent can use.

For example, a skill could allow an agent to:

• Search records.
• Create records.
• Update records.
• Open specific views.
• Create leads.
• Work with website content.
• Generate images.

Sources

Sources provide information the agent can use when responding or completing tasks.

These can include:

• PDFs.
• Web pages.
• Documents.
• Odoo Knowledge articles.
• Other indexed business information.

This means a business can build an agent around a specific purpose rather than simply giving an AI model unrestricted access to the entire organisation.

How Odoo 20 AI Agents Work

A useful way to understand Odoo's approach is to separate the process into four layers:

Instructions → Context → Tools → Action

1. Instructions

The agent is given a role and instructions.

For example:

"You are a sales operations assistant. Review incoming opportunities and assign them according to territory and sales team."

These instructions define how the agent should behave.

2. Context

The agent receives relevant information from the Odoo record or workflow it is working with.

If it is dealing with a sales lead, for example, the relevant lead information becomes part of its context.

3. Tools

The agent is given access to specific tools.

Tools are the functions that allow it to do things rather than merely generate text.

For example:

• Create a lead.
• Update a record.
• Move a document.
• Create an invoice.
• Open a view.

4. Action

The agent determines which tool is appropriate and supplies the necessary information.

The underlying Odoo tool then executes the action.

This separation is important because AI makes the decision, while the configured Odoo tool performs the operation.

AI Agents vs Traditional Automation in Odoo 20

Odoo has supported automated actions for years. So what makes AI-powered automation different?

Traditional automation generally works with predefined conditions.

For example:

If stock quantity falls below 10, send an email.

That works well when the condition is straightforward and predictable.

But many business processes are more difficult to express using simple rules.

Consider an incoming customer email.

The business might want the system to determine:

• What the customer is asking about.
• Which opportunity the email relates to.
• Whether the request is urgent.
• Which salesperson should handle it.
• What action should happen next.

That involves interpretation.

This is where AI can complement traditional automation.

Odoo's AI Server Actions framework allows AI to act as the decision-maker, selecting from predefined tools while the tools themselves carry out the controlled business logic.

This creates a useful model:

Traditional automation:

"If X happens, do Y."

AI-assisted automation:

"Understand what is happening, determine which approved action applies, and then execute it."

Businesses do not have to replace deterministic automation with AI.

Instead, AI can be introduced where human-like interpretation is useful.

What Can Odoo 20 AI Agents Actually Do?

The most important question for businesses is not how impressive AI sounds; it is what the technology can actually do.

Odoo 20's AI capabilities extend into several practical areas.

1. Create and Update Records

AI agents can work with Odoo records using configured tools.

For example, an organisation could build an agent that helps:

• Create leads.
• Create contacts.
• Update customer information.
• Update tasks.
• Create records based on structured instructions.

Imagine a salesperson receiving five new prospects by email.

Instead of manually entering every contact, an AI-assisted workflow could interpret the information and create the relevant records.

The employee can then review the results rather than starting the data-entry process from scratch.

Business impact

This can reduce repetitive administrative work and improve the speed at which information enters the ERP.

2. Automate Sales Processes

Sales is one of the clearest areas where AI agents can have an operational impact.

Odoo itself gives examples such as automatically confirming a sales order when its matching email arrives and assigning leads to the appropriate salesperson.

Consider a typical workflow:

Customer sends an email

↓

AI identifies the customer and context

↓

AI determines the relevant opportunity

↓

AI selects the appropriate action

↓

Odoo updates the record

↓

Salesperson receives the result

Instead of employees manually moving information from email into the ERP, parts of the workflow can happen automatically.

Other potential sales applications

Businesses could explore AI-assisted processes around:

• Lead qualification.
• Lead assignment.
• Opportunity updates.
• Customer follow-up.
• Sales activity summaries.
• Pipeline analysis.
• Record creation.
• Sales reporting.

The exact workflow should depend on the organisation's processes and controls.

3. Automate Lead Management

Odoo 20 also expands AI capabilities around CRM.

Businesses can use Odoo's broader CRM functionality to generate and capture leads, while AI can help interpret and route information.

For example:

New lead arrives

↓

AI reads available information

↓

Identifies industry/location/context

↓

Determines appropriate sales team

↓

Creates or updates the opportunity

↓

Routes the opportunity

This can become particularly valuable for organisations managing multiple sales teams, regions, or business units.

The objective is simple:

Get the right opportunity to the right person without unnecessary manual intervention.

4. Automate Document Processing

Documents are another area where AI agents can be useful.

Odoo's AI framework can work with documents and use their content as context.

For example, an organisation could build a workflow around incoming documents where the agent determines what a document contains and selects an appropriate action.

Odoo's own AI Server Actions documentation provides an example involving documents in the Inbox.

The agent can determine which tool should be used based on document content, while the underlying tools handle actions such as:

• Adding tags.
• Moving documents.
• Renaming documents.
• Creating invoices or bills.

This creates a practical workflow:

Document arrives

↓

AI understands the document

↓

AI determines the appropriate action

↓

Configured Odoo tool executes it

This could reduce manual document sorting and routing.

5. Automate Inventory and Purchasing Decisions

Inventory management is another area where AI can work alongside Odoo's existing automation.

Odoo 20 introduces smarter inventory functionality, including calculations of suggested minimum and maximum stock levels based on recent sales.

The official Odoo release also gives an example of asking an AI agent to review purchasing history and identify vendors that may be worth renegotiating with.

Imagine a purchasing manager asking:

"Which suppliers should we review based on our recent purchasing history?"

Instead of manually opening multiple reports and comparing records, an AI agent can help analyse the relevant information.

Potential applications include:

• Supplier analysis.
• Purchase-history analysis.
• Stock-level monitoring.
• Replenishment workflows.
• Inventory reporting.
• Vendor performance review.

For businesses managing multiple warehouses, stores, or product categories, this can become particularly useful.

6. Use AI to Analyse Business Data

AI agents can also make business information easier to query.

Instead of navigating through several reports, users can ask questions in natural language.

For example:

"Show me our open sales opportunities by salesperson."

Or:

"How much did we invoice last quarter compared with this quarter?"

Or:

"Which products on our manufacturing schedule have no forecast data?"

Odoo's MCP documentation gives examples of this kind of natural-language analysis across sales, manufacturing, inventory, and financial information.

This is important because many employees know what question they want answered but do not necessarily know which report, filter, or menu they need to use.

AI can become an interface between the employee and the business data.

7. Give Finance Teams an AI Assistant

Odoo 20 extends AI into Accounting as well.

Odoo's official release highlights an Accounting AI agent that can answer questions about financial information, including receivables and cash flow.

It can also assist with reconciliation analysis and provide summaries and recommendations.

For finance teams, this could make financial investigation more conversational.

Instead of:

Open report → apply filters → export → analyse → repeat

the interaction can become:

Ask → retrieve → analyse → investigate

This does not remove the need for accountants or financial controls.

It can instead reduce the amount of time spent navigating information.

Possible finance use cases

• Receivables analysis.
• Cash-flow questions.
• Reconciliation investigation.
• Financial summaries.
• Report analysis.
• Identifying anomalies that require review.

The accountant remains responsible for financial decisions and controls.

AI becomes another tool for accessing and analysing the information.

8. Build Agents Around Company Knowledge

AI is only as useful as the information available to it.

A generic AI model may know a great deal about the world, but it does not automatically know:

• Your company's policies.
• Your internal procedures.
• Your product specifications.
• Your HR policies.
• Your customer-service guidelines.
• Your approval processes.

Odoo 20 allows AI agents to use Sources such as documents, PDFs, web links, and Knowledge articles.

For example, a business could create an internal support agent with access to:

• Employee handbook.
• Procurement policy.
• Expense policy.
• Product manuals.
• Customer-service procedures.
• Internal FAQs.

An employee could then ask:

"What is the approval process for this expense?"

The agent can use the configured sources to provide an answer.

Businesses can also restrict an agent to its designated sources when that is appropriate.

9. Use AI to Improve Website Work

AI agents are not restricted to back-office processes.

Odoo 20 also brings AI into website creation and management.

Odoo says users can describe the page they want and have Odoo create it.

AI can also assist with:

• Website copy.
• Promotions.
• Event pages.
• Product pages.
• Website snippets.
• Translations.
• SEO optimisation.
• Interactive website elements.

For marketing teams, this changes the workflow from:

Create everything manually

to:

Describe → Generate → Review → Refine

The human still plays an important role.

AI can create the first version, but marketing teams should review accuracy, tone, branding, SEO intent, and customer relevance before publishing.

10. Connect External AI Tools to Odoo Through MCP

One of the most significant technical additions in Odoo 20 is its Model Context Protocol (MCP) server.

MCP is an open standard that allows AI systems to interact with external applications and data.

Odoo 20 provides an MCP server that can allow compatible external AI clients to interact with an Odoo database.

This means Odoo does not have to be isolated from the AI tools a business already uses.

Depending on how the administrator configures the available tools and permissions, an external AI agent can:

• Read Odoo information.
• Search records.
• Analyse data.
• Create records.
• Update records.
• Interact with configured Odoo tools.

For example, a business could connect a compatible external AI workflow and ask:

"Show me our open opportunities grouped by stage and salesperson."

Or:

"Create contacts for these new leads."

Or:

"Check whether this customer has a tax ID and update it if necessary."

The exact capabilities depend on which tools the organisation exposes.

How Odoo 20 MCP Works

The MCP workflow can be simplified to:

External AI client

↓

Odoo MCP server

↓

Authenticated Odoo user

↓

Approved tools

↓

Odoo database

The external AI client sends a request.

Odoo authenticates the user and determines what the connected user is allowed to access.

The AI client can then invoke tools that have been exposed to it.

Odoo's documentation states that several read-oriented tools are available by default, while other tools must be explicitly exposed by an administrator.

This is an important security principle:

Connecting an AI tool to Odoo does not automatically mean giving it unrestricted access to the database.

Administrators decide which tools are available.

Why Access Control Matters With AI Agents

AI automation introduces a new consideration for ERP security.

Traditional ERP permissions control what users can access and modify.

With AI agents, businesses also need to consider:

What is the AI allowed to do on behalf of the user?

Odoo's AI architecture provides mechanisms for controlling this.

For example:

• Agents can be assigned specific skills.
• Skills can have specific tools.
• Tools can enforce business rules.
• MCP tools can be selectively exposed.
• Read-only tools can be identified.
• User permissions still matter.

This is why AI implementation should not simply be:

"Give the AI access and see what happens."

A better approach is:

Define the task → define the data → define the tools → define permissions → test → monitor → expand.

AI Should Not Replace Business Rules

This is one of the most important points businesses should understand about Odoo 20 AI.

AI is useful for interpretation and decision-making, but deterministic business rules should remain deterministic wherever possible.

Odoo's documentation makes this distinction clear in its AI Server Actions architecture.

The AI component decides which tool to use.

The tool contains the actual execution logic and should enforce the necessary business rules.

For example:

AI:

"This looks like a high-priority customer request."

Business rule:

"Only users with the appropriate permission can change the ticket priority."

The AI can identify the situation, but the underlying business logic should still protect the organisation.

This architecture is important because AI can make mistakes.

The goal should therefore not be:

"Let AI do everything."

It should be:

"Let AI handle the parts of the process where interpretation adds value, while keeping critical controls deterministic."

Five Practical Odoo 20 AI Workflows Businesses Can Consider

To understand the potential better, consider five common business situations.

Workflow 1: Sales Order Processing

Before

Customer sends an email.​

Salesperson opens email.

Salesperson searches for customer.

Salesperson finds the opportunity.

Salesperson reviews the request.

Salesperson updates Odoo.

Salesperson confirms the order.

With AI-assisted automation

Customer sends an email.

↓

AI identifies the matching customer/order context.

↓

AI determines the appropriate action.

↓

Configured Odoo tool executes the approved process.

↓

Salesperson reviews the result where required.

Potential benefit: Less repetitive administrative work.

Workflow 2: Lead Assignment

Before

Marketing generates leads.

A sales manager reviews them.

The manager determines who should receive each lead.

The manager assigns them manually.

With AI

A new lead enters Odoo.

↓

AI analyses available information.

↓

AI identifies the appropriate team or salesperson.

↓

Lead is routed automatically.

Potential benefit: Faster lead distribution and less manual routing.

Workflow 3: Document Processing

Before

A document enters the company.

An employee opens it.

Reads it.

Determines its type.

Adds a tag.

Moves it to the appropriate folder.

Possibly creates another record.

With AI

Document arrives.

↓

AI analyses its content.

↓

AI chooses the appropriate approved tool.

↓

Odoo executes the configured action.

Potential benefit: Less manual document administration.

Workflow 4: Inventory Monitoring

Before

Purchasing team checks inventory.

Reviews recent sales.

Checks supplier lead times.

Determines what needs replenishment.

Creates purchase orders.

With Odoo 20

Odoo analyses stock and recent sales.

↓

AI or automated workflows identify relevant situations.

↓

Suggested action is generated.

↓

The purchase process can be initiated according to configured rules.

Potential benefit: More proactive inventory management.

Workflow 5: Financial Analysis

Before

A manager asks:

"What is our outstanding receivables position?"

The finance team runs reports.

Filters them.

Exports data.

Analyses the results.

Responds.

With AI

Manager asks the question.

↓

AI retrieves the relevant financial information.

↓

AI provides an explanation based on the available data.

↓

Finance team investigates or validates where necessary.

Potential benefit: Faster access to financial information.

What Are the Benefits of Odoo 20 AI for Businesses?

The potential value of AI agents goes beyond simply "saving time."

1. Less repetitive work

Employees can spend less time on routine data entry, record updates, and administrative processes.

2. Faster response times

Automated workflows can respond to events as they happen instead of waiting for an employee to notice them.

3. Better access to information

Employees can ask questions in natural language instead of learning where every report and filter is located.

4. More consistent processes

Once an AI workflow is properly configured, the same process can be applied repeatedly.

5. Better use of employee time

Employees can focus more on tasks that require judgement, communication, relationship management, and decision-making.

6. Connected automation

Because AI operates inside the ERP, it can work with the same records and workflows already used by the business.

What Odoo 20 AI Does Not Mean

There are several misconceptions businesses should avoid.

AI does not mean your ERP can run without people.

Human oversight remains important, especially for sensitive or high-impact processes.

AI does not automatically understand your business.

Agents need appropriate instructions, context, sources, and tools.

AI does not eliminate implementation.

In fact, the opposite may be true.

The more deeply AI interacts with business processes, the more important it becomes to understand those processes properly.

AI does not make poor processes good.

If the underlying process is inefficient, automating it can simply make the inefficient process happen faster.

What Businesses Should Prepare Before Implementing Odoo 20 AI

Businesses considering Odoo 20 should not begin by asking:

"Where can we put AI?"

Start with the processes.

Step 1: Identify repetitive processes

Look for tasks employees perform repeatedly.

Examples:

• Assigning leads.
• Sorting documents.
• Updating records.
• Sending routine notifications.
• Checking stock.
• Preparing reports.

Step 2: Identify decision-heavy processes

Some processes cannot be handled by simple rules because they require interpretation.

These may be strong candidates for AI.

Step 3: Define the business rules

Before allowing AI to take action, determine:

• What can it do?
• What cannot it do?
• What requires approval?
• What requires human review?
• What permissions apply?

Step 4: Prepare your data

AI depends on the quality of the information it receives.

If customer records are incomplete, product data is inconsistent, or processes are poorly documented, AI automation will be less reliable.

Step 5: Start with controlled use cases

Don't automate 20 processes at once.

Choose one or two processes where the benefit is clear, and the risk is manageable.

Test them.

Measure them.

Improve them.

Then expand.

How Much Human Oversight Should AI Have?

Not every process should have the same level of automation.

A useful approach is to divide processes into three categories.

Level 1: AI assists

AI provides information or recommendations.

A human makes the decision.

Example:

"Which suppliers should we review?"

Level 2: AI prepares

AI creates a proposed action.

Human reviews and approves.

Example:

Draft a purchase order based on identified replenishment needs.

Level 3: AI executes

AI can perform the action automatically within defined rules.

Example:

Assign new leads to the correct salesperson according to established criteria.

The appropriate level depends on the risk and importance of the process.

For financial transactions, employee records, customer commitments, and other sensitive operations, businesses may prefer stronger approval controls.

Odoo 20 AI and Business Process Improvement

This is where AI and ERP implementation come together.

AI agents are powerful, but they do not replace business process analysis.

Before automating a process, businesses should understand:

What happens today?

Who performs each step?

What information is required?

Where do delays occur?

Which steps are repetitive?

Which decisions require human judgement?

What should happen automatically?

Where should approval remain?

Only after answering these questions can an organisation determine where AI fits.

For this reason, businesses implementing Odoo 20 should consider AI as part of a broader business process improvement exercise, not as an isolated feature.

Odoo 20 AI for Different Types of Businesses

The opportunities will differ depending on the organisation.

Manufacturing

AI can assist with:

• Production information.
• Inventory analysis.
• Supplier analysis.
• Manufacturing planning.
• Document processing.
• Workflow automation.

Retail

Potential use cases include:

• Inventory monitoring.
• Customer data.
• Sales analysis.
• Lead management.
• Purchasing.
• eCommerce processes.

Professional Services

AI can support:

• Lead assignment.
• Project creation.
• Document processing.
• Timesheet-related workflows.
• Customer communication.
• Reporting.

Distribution

Potential applications include:

• Stock monitoring.
• Purchasing.
• Supplier analysis.
• Sales order processing.
• Customer management.

Healthcare and membership organisations

AI can potentially assist with:

• Document workflows.
• Customer/member communication.
• Record management.
• Internal knowledge.
• Administrative automation.

The exact implementation should always be based on the organisation's processes, data and regulatory requirements.

Odoo 20 AI and the Future of ERP

Odoo 20 reflects a broader change in enterprise software.

ERP systems traditionally required employees to learn how to navigate the software.

AI introduces another possibility:

Employees can increasingly tell the system what they want in natural language.

Instead of knowing:

Which menu contains this report?

the employee can ask:

"Show me our unpaid invoices over 30 days."

Instead of manually navigating a workflow:

"Create a project for this signed contract."

Instead of manually sorting information:

"Review these documents and classify them according to our configured categories."

This does not mean traditional interfaces disappear.

It means ERP systems can increasingly offer conversation as another way of interacting with business processes.

Odoo 20 AI: What Businesses Should Consider Before Adopting It

AI should be evaluated alongside the same questions businesses ask when implementing any major ERP capability.

Business fit

Does the AI capability solve a real operational problem?

Data quality

Is the information AI will use accurate and structured?

Process maturity

Is the process already clearly defined?

Security

What data can the AI access?

Permissions

What actions can the AI perform?

Human oversight

Which actions require approval?

Training

Do employees understand how to work with AI-assisted workflows?

Measurement

How will the business determine whether the automation is actually producing value?

These questions are important because successful AI adoption is not simply a technology project.

It is a business process project.

The Real Opportunity With Odoo 20 AI

The most interesting thing about Odoo 20 AI is not that businesses can "chat with AI."

It is that AI is becoming part of the operational layer of the ERP.

Sales, purchasing, inventory, accounting, documents, websites, CRM, and other business processes can increasingly incorporate AI-assisted decision-making and automation.

That creates a new model for business software:

People + Processes + ERP Data + AI

When these components are properly connected, employees can spend less time moving information between systems and more time acting on the information.

But achieving that outcome requires more than switching on an AI feature.

The business needs to identify the right processes, establish the right rules, prepare the right data, and configure the right tools.

Should Your Business Start Using Odoo 20 AI?

If your organisation is considering Odoo 20, don't start with the question:

"How many AI features can we implement?"

Start with:

"Which parts of our business are consuming too much time because people are repeatedly reading, deciding, entering, checking, or moving information?"

Those are the processes worth investigating.

Odoo 20 gives businesses a new set of tools for addressing them.

AI agents can interpret requests, use business context, retrieve information, call approved tools, and perform configured actions.

The opportunity is therefore not simply to add AI to your ERP.

It is to rethink how work moves through your organisation.

Conclusion: Odoo 20 Makes AI Part of the Workflow

Odoo 20 represents an important development in the evolution of ERP software.

AI is no longer limited to generating text, summarising information or answering questions.

With AI agents, Skills, Sources, AI Server Actions and the Odoo MCP server, organisations have new ways to connect AI with actual business data and processes.

That opens the door to workflows where:

An event happens → AI interprets it → the appropriate tool is selected → Odoo performs the action → the result is recorded.

For a business dealing with hundreds or thousands of repetitive processes every day, that can create meaningful opportunities to improve efficiency.

But successful implementation will depend on something that existed long before AI:

A clear understanding of how the business works.

The most valuable Odoo 20 AI implementation will not necessarily be the one with the most AI agents.

It will be the one that applies AI to the right processes, with the right controls, using the right business data.

Ready to Explore Odoo 20 for Your Business?

If you are considering Odoo 20 and want to understand how its AI capabilities could fit into your organisation, the first step is to assess your current processes.

At erpSOFTapp, we help businesses understand their processes, identify opportunities for improvement, and configure Odoo around the way they actually operate.

Whether your goal is to automate sales administration, improve inventory processes, connect finance and operations, streamline document management, or explore AI-powered workflows, the right implementation starts with understanding the business problem first.

Don't implement AI because it's new. Implement it where it can make your business work better.

Talk to us at erpSOFTapp about implementing Odoo 20 for your business.