AI by industry

AI for solar installers in Portugal

Portuguese solar installers rarely lose deals on price. They lose them on the four days it takes to get a quote back out.

Rooftop solar in Portugal has moved from a niche to a queue. Falling panel prices, reduced VAT on equipment, and electricity bills that homeowners now watch monthly have produced more inbound enquiries than most installers can process. The constraint is no longer demand. It is the gap between an enquiry arriving and a priced, credible proposal landing in the customer's inbox, because in that gap two competitors also quote.

Where the days actually disappear

• Unqualified leads. A large share of enquiries are renters, apartments without roof rights, or roofs facing the wrong way. Every one of them still eats a phone call and sometimes a site visit.
• Quote assembly. Roof area, orientation, shading, consumption from the last twelve bills, panel and inverter selection, then the same document rebuilt by hand for each customer.
• Registration and paperwork. UPAC registration, DGEG formalities, grid operator steps, and the certificates the customer asks for months later. All of it is deadline-bound and none of it is billable.
• After the install. Monitoring alerts, performance questions, warranty claims, and the annual service reminder that quietly never gets sent.

What an agent does about it

The highest-return first project is almost always intake. A WhatsApp and web agent asks the five questions that decide whether a roof is worth visiting, in Portuguese or English, at the moment the customer is interested rather than three days later. It reads the uploaded electricity bill, pulls out annual consumption and tariff, and either books a survey in the installer's calendar or explains politely why this address is not a fit.

Behind that, a quoting workflow assembles the proposal from the surveyor's inputs and your own price list, so the document goes out the same day. Post-install, a monitoring agent watches inverter output against expected yield and raises a job when a string underperforms, which is how service revenue gets found instead of waited for.

What we do not automate

Nobody should be generating engineering sign-off, structural judgement, or regulatory declarations from a language model. The agent handles intake, drafting, chasing, and monitoring. A human stays on the technical decisions and signs the documents that carry liability. That line is where these projects either build trust with a technical team or lose it in the first month.

Digiton builds and runs these systems rather than handing over a tool and leaving. If you want to know which of your steps would return the most hours, a short AI audit maps them against your actual job flow before anything gets built.

Frequently asked questions

Related