Odoo 20 for maintenance businesses: less rekeying, more answers

The Odoo update released in late September 2026 expands the AI capabilities in the management system: agents can create data from an instruction file. They can also update existing data, run within workflows, and request permissions you can grant at the click of a button. For your maintenance business, these capabilities can help reduce rekeying in the office and make information easier for technicians to access. The features of Field Service, the app for organising service jobs in the field, have moved into Planning, the app for scheduling shifts. Before enabling agents, review your permissions: when an external agent connects to Odoo using an access key, its access depends on the identity and permissions of the user who authorises it.

This article explains what changes in practice, what to check before upgrading and where to start.

Where maintenance businesses lose time today

In a typical maintenance business with around ten technicians, some of your office team's work involves entering information into the system that already exists elsewhere. A customer sends a document listing the systems that need maintenance, and someone rekeys it line by line. A technician calls to ask whether an air-conditioning unit is covered by the contract, and the person answering opens three screens before they can say. A service report arrives that evening, written in a rush, and needs to be read through and entered into the system the next morning. It's a familiar scenario, used here to illustrate the problem.

You need that information to schedule jobs, prepare invoices and piece together what has happened to a system. Transferring it manually takes time, even when it is already written in a document. Extracting information from text is already a common use of AI: in ISTAT's “Enterprises and ICT – 2025” survey, 70.8% of Italian businesses with at least ten persons employed that use AI use it for this purpose. That helps explain why documents are a useful place to start.

What agents do in Odoo

The official release notes describe a step forward from the previous version, which already let you ask the system questions and assign tasks to agents. With this update, you can ask an AI agent to create data from an instruction file, such as a PDF, or update existing data. Agents can also be started by automated or scheduled actions, allowing their work to fit into your business processes without you having to write a new request each time.

For your technicians and office team, how you interact with the agent matters too. You can dictate requests and ask questions about a file you are viewing, such as an equipment manual. As agents work, they show what they are doing and let you answer questions and approve permission requests by pressing a button. Dictation applies to requests you make to the agent: the notes do not describe a dedicated feature for completing service reports in the field.

Odoo also lets you connect external agents to your system. When you connect an agent using an access key, that key carries the identity and permissions of the user who authorises it. This determines which data the agent can access and which actions it can perform. To revoke that access, delete the key.

In the illustrative opening scenario, someone in your office uploads the customer's document as an instruction file and asks the agent to create records for the systems listed. Once the agent has finished, that person compares the data it has created against the document and corrects any errors before using it for work. This illustrates how the agents' general capabilities could be used, with a person checking the results.

Field Service moves into Planning: what your technicians can access on a job

If you run a service business, one practical change concerns how the apps are organised. The release notes state that the Field Service app has been discontinued and its features are now integrated into Planning.

Your technicians can access documentation from their shift screen. Planning lets you track service jobs and contract coverage, with coverage managed through Subscriptions. Automatic scheduling takes customer locations into account. Your office team can track technicians on a live map, collect requests through a form on your website and bill for travel by distance or charge a fixed fee for travel time. You can also collect customer ratings for service jobs. In Helpdesk, the app for managing support requests, AI identifies similar requests and suggests responses to the team.

If you already use Field Service, the move to this release needs preparation: the app you use to organise service jobs is changing. Before upgrading, walk through your workflow from the customer's request to closing the job, and identify which steps you could simplify. Recreating every existing habit in the new setup would also carry over unnecessary steps.

An Italian example: agents handle requests, people stay in control

You can already see this approach in Italian companies with field operations. In early November 2026, Bludigit, the Italgas group's IT company, described a system that uses multiple AI agents, supervised by people, to handle requests and complaints. Within the initial test scope, overall handling time for requests and complaints fell by 60%. A Databricks case study on Italgas describes how complaint handling in 2023 was fragmented and manual, forcing staff to move between systems.

The scale and resources are different from those of a small maintenance business. But you may recognise the starting problem even if your office has just two people: the information exists, but it is scattered and has to be brought together before anyone can deal with the request. That is the point to focus on in the Italgas example. In a smaller business, you can keep the work assigned to AI narrowly defined while people remain in control of what it prepares.

Before enabling agents: permissions, approvals and responsibility

An external agent connected to Odoo through an access key works with the identity and permissions of the user who authorises it. If that user has overly broad permissions, or permissions that were never revoked, the agent can access that data too. Permissions are also an issue in ShareGate's survey of 851 people: 29% report that an AI tool surfaced sensitive internal data that should not have been accessible.

Before enabling agents, establish who can see what, including external contractors and portal users. Decide which actions the agent can perform and which need confirmation: viewing a record and changing it have different consequences. For source material, you need to know who keeps it up to date and who removes outdated documents. An old procedure can lead to a wrong answer. In 2024, a Canadian court held Air Canada responsible for incorrect information provided by the assistant on its website.

In Italy, Article 11 of Law 132 of 2025 requires employers to inform workers about the use of AI in the circumstances and in the manner set out in employment law. There is also an operational factor to assess: with this release, AI features require credits that you can purchase through the built-in service. Include credit usage when calculating the cost of the process you want to automate.

Where to start

You can start with a single step: the one that currently involves the most rekeying or the most calls between technicians and the office. Review the lists of systems your customers send you, your service reports and the requests you receive by email. To choose where to start, look at completed work and find where time is being lost.

With Gitogi, you start with a month of completed work: that month's service reports, invoice line items and service contracts. Together, we look for missing links between service jobs, contract coverage and invoices. We identify completed jobs still waiting to be checked or invoiced, hours and materials entered more than once, work covered by the contract and work still to be priced. We also look at how long it takes to issue an invoice after a job is finished. Then we measure how much time or margin you lose at the most costly step. Before proposing any development work, we show you the numbers: costs, estimated benefits with the assumptions clearly stated, and payback periods. If development work does not make financial sense for that step, we tell you.

We are an Odoo Partner and work with businesses, starting with how their technicians and office teams actually work. When we develop custom software, we connect it to the management system your business already uses, and the code developed specifically for you remains yours. Odoo is an option if you are looking for Odoo or want to change management systems. You can read about how we work with maintenance businesses at https://www.gitogi.com/en/field-service/ and how we tailor AI integration to your business's needs at https://www.gitogi.com/en/servizi/integrazione-ai/.

If you are considering Odoo, or someone in your office is still rekeying documents by hand, write to us and tell us how you work. We will start with a month of completed work, using your service reports, invoice line items and service contracts.