A system for a company that installs and services air conditioning and heat pumps. It keeps the service calendar so nobody has to, lays out a week of technician work, and ties each sale to the specific unit standing at the customer.
Captures from a running instance; the company in them is invented
Several hundred customers, one to a handful of units each, every unit on its own service cycle.
A split system once a year, heat recovery every six months, a shop installation every six. On top of that, breakdowns that have to be wedged between planned visits, and enquiries for new installations arriving all through the season.
A spreadsheet holds that to about a hundred units. After that it turns into calling back customers who remembered their own service date first, and searching an inbox for who was on site last and what they did. The customer does not leave because the work was bad. They leave because nobody got in touch.
The whole system rests on one loop, and the remaining screens only serve it. The order below is not decoration - it is the actual flow of the work.
The address the equipment stands at, with coordinates, a technical contact and access notes. One customer can have several.
Manufacturer, model, serial number, install date, warranty, and its own service interval counted in months.
The system works out the next service date from the last visit. Once it passes, the unit appears on the coordinator’s worklist without anybody putting it there.
A service, a breakdown or an installation goes to a technician, with a date, a time budget and the customer’s own description of the problem.
A finished service writes the work done into the unit’s history and resets the clock for the next cycle. The loop closes itself.
Jobs move through four columns. A technician sees their own, the coordinator sees the whole team, and anything overdue turns red on its own instead of waiting to be noticed.
A component of our own on open map tiles - no paid keys, no modules from the paid edition of Odoo. Points are coloured by how overdue the service is; the filters at the top switch the view in one click.
Who was there, when, and what they did is readable from the unit itself. No reconstructing it from memory, no digging through mail from a year ago.
Enquiries did not get a model of their own. They extend the standard Odoo pipeline with a site, a unit and a case type, so all the stage and quotation mechanics work the way they do everywhere else - and the company gets the link to a specific installation on top.
The calls that weigh more in a project like this than the number of screens does.
The pipeline is an extended crm.lead, not an opportunity model of our own. Less code to maintain, full compatibility with the rest of Odoo, and free updates to mechanics nobody will build better than the vendor.
The product installs at a customer without a single spare record, while the demo module adds an entire fictional company for a walkthrough. A script with a fixed random seed generates the set, so the demo reproduces identically from one command.
The map library sits in the repository, not on somebody else’s CDN. The component fetches sites in one query and works out service urgency itself, so the map runs on the free edition of Odoo with no key to a paid geocoder.
The rules keep a technician to their own jobs and let them read records without editing them. The matrix has a test suite of its own, because it is the one thing you cannot see on screen until it is too late.
Odoo 19 Community and Postgres in containers, with the product module kept separate from the demo-data one.
Twenty-three unit tests cover the service cycle, searching on computed fields, and the permission matrix. On every change to the repository the installation starts from nothing, the suite runs in full, and a separate step checks that the demo data still matches the script that generates it.
The delivered interface is in Polish, field labels and status names included - the crew using it works in Polish, and a technician should not need a translation to read their own worklist. That is what the captures on this page show. The same build ships in whatever language the team on site actually speaks.
The company in these captures, its customers, addresses, serial numbers and staff names are invented. A script generates the whole data set from a word list, and any resemblance to a real company would be coincidental. Manufacturers and model names are real, because without them the records would not look believable.
A few sentences on what is in the spreadsheet today and who it is hurting. We answer with scope, a timeline and a quote within 24 hours.
Free consultationConstruction digitises more slowly than other sectors, and its problems are countable: budgets overrun by 20-50% and...
On the factory floor AI watches three things: process parameters, machine condition and product quality.
In logistics AI goes where the scale is: routes, stock levels and shipping paperwork.
Salesforce is the most widely used CRM in the world, with Sales Cloud, Service Cloud, Marketing Cloud and Commerce...
HubSpot is a CRM and marketing automation platform made of Sales Hub, Marketing Hub, Service Hub and CMS Hub.
Pick the type of project, the features and the scale, and the calculator works out the effort and the schedule: a web...