--- title: "AI for maintenance and HVAC companies · Quentin Andrieux" description: "Service reports, nameplates, follow-up quotes: AI for maintenance, HVAC and heating companies. A prototype running on your data in five days." url: https://quentinandrieux.com/en/maintenance/ lang: en author: Quentin Andrieux updated: 2026-11-06 --- > Version texte de https://quentinandrieux.com/en/maintenance/ (the HTML page is the reference). Author : Quentin Andrieux, https://quentinandrieux.com/ Maintenance · HVAC · heating # Your service reports and quotes, ready the same day. AI for maintenance, HVAC and heating companies. One day on site with your technicians and your office, then a tool running on your data in five days. Your teams keep working the way they do: the tool adapts to them. [Get a written plan](https://quentinandrieux.com/en/maintenance/#plan) [Book 30 minutes](https://cal.com/quentin-andrieux/30min) For business owners who want to see something work before they decide. Fictional example · one job Technician's photos Nameplate 14:05 Controls 14:11 Pump 14:19 That evening, at the office Service report to sign Follow-up quote to send What the tool does with your documents ## The paperwork that slows your jobs down, handled by AI. What your technicians already produce, turned into documents ready to send. 01 ### Service report The technician dictates a note or takes a photo. A standard report goes to the customer and the office, with no re-typing. 02 ### Nameplate Read from the technician's photo: brand, model, serial number. The equipment is found or created in your database. 03 ### Follow-up quote Built from the findings, your prices and your parts catalogue. Missing items are flagged before it goes out. 04 ### Site survey report Generated from the photos of the visit: equipment inventory, condition, points to watch. 05 ### Maintenance contract Pre-filled from the equipment found on site, with the matching services. 06 ### Breakdown requests Emails, forms, calls: each request is read, qualified and routed to the right technician or person. Plugged into your tools ## Nothing to replace, nothing to migrate. The tool reads your data where it already lives. The prototype changes nothing in your systems; writing directly into your software comes at the next step, once your teams have validated the tool. Field service software ERP In-house Access software Spreadsheets Email Phone photos Delivered · anonymised engagements ## Tools running at field-based companies. A business owner Have you actually built this before? Quentin Yes. Here are two examples, from field service companies. **÷2** time spent per service call, as estimated by the team ✓ Delivered HVAC company · new owner The owner had just taken over the company. Interviews with the teams, then an app built piece by piece on their real processes: service reports generated automatically, nameplates read by AI. 2-week sprint AI image reading **÷2** maintenance intervention time, as estimated by the team ✓ Delivered Energy manufacturer · maintenance division A mobile app to bring field information back to the office and manage it. Built with the teams, on their real data. Prototype in 6 days Works on mobile Your teams stay in control ## AI prepares, your teams decide. ### Nothing is guessed An unreadable reference, an unknown unit, a missing price: anything the tool is not sure about goes to a person for validation. ### Everything is logged Every action of the tool is kept: what it read, what it produced, who validated it. ### Your data stays yours Access limited to what is needed, a confidentiality agreement, and everything built belongs to you. ### Every correction counts When one of your people corrects the tool, the correction becomes a rule it applies from then on. The method ## Monday on site, Friday it runs. 1. 01 Monday ### Observe One day at your company, with the technicians and the office, to see how the work is really done. 2. 02 Tuesday → Friday ### Build A prototype on your real data, around the document that costs you the most time. 3. 03 The following weeks ### Go live The tool goes into your teams' hands, with human validation of unusual cases. 4. 04 Next document ### Move to the next Once the first document is under control, we take on the next one with the same method. Free · no commitment ## A written plan, for your company. Tell me which documents slow you down. I'll send back the two or three worth automating, and the shortest path to a tool that runs. Thank you, received. I'll reply within two business days. Rather talk it through? [Book 30 minutes](https://cal.com/quentin-andrieux/30min) Frequently asked ## What people ask me before we start. ### Will my technicians have to change their habits? No. They keep dictating, taking photos and filling things in as they do today. The day on site is there to understand how they work, so the tool adapts to it. ### Does it work with our current software? The prototype reads your data where it is: field service software, ERP, spreadsheets, email. If your software has no ready-made connection, we use its exports. Writing directly into your systems is built afterwards, with your vendor if needed. ### What happens when the AI gets it wrong? Anything it is not sure about goes to a person for validation instead of being guessed. Each correction is then used to improve the tool. ### How long before it runs? The diagnostic takes five business days, one of them at your company, and ends with a prototype running on your data. Putting a first document into production then takes a few weeks, depending on your systems. ### What does it cost? The five-day diagnostic, prototype included, costs €5,000 excl. VAT (indicative price). What comes next depends on the document chosen and your systems: it is costed in the roadmap you receive at the end of the diagnostic.