Services
Solutions
Use cases
Industries
Tools
Pricing
Book a call
Industry — Trades

AI in the trades — construction, HVAC, electrical

What really takes the load off, what isn't worth it, and where people typically start. Five concrete from daily work, honest trade-offs, and notes on funding.

The daily reality AI has to fit into

Phone call before seven in the morning. Site measurement on the phone at the construction site. Material planning during the lunch break. Quote in the evening at the kitchen table — three hours for a job the customer then leaves open for two weeks. Sunday morning for the bookkeeping, because there was no time during the week. A sick call at six-thirty shifts the whole day's planning.

In a typical trades business with ten to thirty employees, admin eats up roughly a third of the working time — for the owner and the site lead, often more. isn't a new employee. It's a set of tools that makes the recurring office work easier — the work that eats into evenings and weekends.

What matters: honestly sorting out what's worth it and what isn't. replaces neither the master craftsperson nor the site lead. It prepares, suggests, classifies — a human edits and signs off. That dividing line decides whether takes hold in the business or stays stuck in trial mode.

Five places where AI in the trades really makes sense

For each case: the situation today, where plugs in, a pointer to the matching setup tiers, and an honest trade-off.

01. Quote from on-site measurements, photos and voice notes

Situation today

Thirty minutes of measuring on site, then two to three hours in the evening at the kitchen table — photos in the gallery, handwritten notes on a slip, material prices and hourly rates somewhere between head and Excel. Four out of ten quotes end up with the competitor in the end, because they go out too slowly.

Where AI helps

structures the voice note and photos into a draft — line items, hours, material suggestion from your own standard clauses and price lists. You edit and sign off, instead of starting from a blank page.

What it can't do

Statics, connection requirements and safety-relevant quantities remain a master-craft job. delivers a draft, not the binding quote.

02. Searching VDE standards, data sheets and connection diagrams

Situation today

VDE 0100, the connection diagram for the old Stiebel module from 2018, the maintenance interval from the manufacturer's data sheet, your own standard text for connection conditions — stored somewhere, but where? Half an hour of searching per day adds up.

Where AI helps

Your own knowledge base with standards, data sheets, connection diagrams and your standard clauses, semantically searchable. “How did we handle the connection with Stadtwerke Hannover back then” finds the right note even without the exact keyword.

What it can't do

Lives on maintenance. If you don't feed in sources, you have a nice search over nothing. Recommendation: one hour of upkeep per quarter, otherwise the base goes stale.

03. Classify incoming mails, prepare appointments

Situation today

Thirty mails per day in the office: requests for quotes, appointment requests, complaints, supplier replies, one reminder and three newsletters. An office assistant sorts for an hour before anything actually gets done — and the callback that was supposed to happen today disappears in the stack.

Where AI helps

classifies incoming mails (request, appointment, complaint, supplier), suggests a reply and books appointment requests into the calendar with capacity check. Sending only after human approval.

What it can't do

No auto-send — otherwise an appointment in the Lüneburg Heath lands with a Hannover business. prepares, a human signs off.

04. Keeping maintenance contracts and material movement in sight

Situation today

Two hundred maintenance contracts, renewal dates in an Excel that only one person maintains. Material stock in the head of the lead warehouse fitter. If they're sick, three days are spent guessing.

Where AI helps

Maintenance deadlines are watched and presented with contract context for confirmation. Delivery notes via into the material inventory, minimum-stock alerts instead of gut feel. Master data stays maintained because the system reports gaps.

What it can't do

Stands and falls with maintained master data. Anyone who feeds in wrong contract data automates wrong reminders — faster than by hand.

05. Defect report from the site's photo documentation

Situation today

The fitter photographs the site, the site lead writes the defect report in the office in the evening — pieced together from memory and the photo gallery. With three sites a day, it turns into a Sunday-evening task.

Where AI helps

The fitter's photo plus a voice note become a structured defect report: item, description, suggestion for rework. The site lead reviews and adds, instead of writing from scratch.

What it can't do

Safety-relevant assessments — VDE , structural suitability, fire safety — remain a master-craft job. The draft is preparation, not sign-off.

What in the trades isn't (yet) worth it

Four promises that, in trades businesses of this size, usually cost more money than they save — as of 2026:

Generative design / CAD automation

AI-assisted design is a very specialised market in 2026 with its own tools — for a typical construction, HVAC or electrical business with ten to thirty employees, the payoff is rare. Classic CAD and measurement software remains the better lever.

Predictive maintenance on your own machinery

Requires sensor history over years. With few machines and no structured data, the cost/benefit is negative. Supplier and manufacturer service offerings are often more viable here.

AI as a structural or standards reviewer

Liability and professional law set clear limits. On safety-relevant assessments, may suggest, not decide — and the effort of reviewing suggestions can be higher than doing the assessment itself.

Fully automated dispatching

Real dispatch knows the Hans who has no phone today, and the customer who only takes Thursdays off. Classic dispatch tools with an briefing are often the more honest choice than auto-plans.

What needs to be thought through for AI in the trades

Four legal points that belong on the table in every project in the trades — not as a brake, but as a setup prerequisite:

GoBD for invoices and receipts

Immutability, retention periods, traceability — automated workflows have to comply. Delivery notes, incoming and outgoing invoices stay audit-proof, the audit trail is mandatory.

ZUGFeRD and X-Rechnung

In B2B with the public sector mandatory since 2025/2026, in B2B invoices gradually the standard. AI-supported receipt pipelines have to produce or read the format correctly — usually via your own accounting or industry software.

People in construction-site photos

Site photos can be used, but people in them are data-protection relevant. With photo-doc pipelines, facial recognition or anonymisation is planned in — or the shots are filtered before processing.

AI literacy under Art. 4 EU AI Act

Since February 2026, all businesses are required to train staff in the proper use of — including trades businesses with few tools. A pragmatic, annually refreshed training meets this duty.

Tools that often already run in trades businesses

doesn't replace these systems — it plugs in. Where the interface typically sits:

Industry software

BauSU, FORMAT-Handwerker, Streit V.1, Pace, Sage Handwerk, addisonOne Handwerk — typically plugs in via the export interface or by email hook, not via deep integration

Accounting & receipts

DATEV, sevDesk, Lexware, Buchhaltungsbutler — via certified interfaces or DATEV receipt transfer, depending on the tax advisor

Measurement & photo docs

Sander & Doll, magicplan, PlanRadar, BauMaster, Capmo — photo and voice-note material from these apps can be fed into an pipeline

Material and warehouse management

Mareon, plancal nova, in-house solutions — typical plug-in point: CSV or export for stock reconciliation

How people typically get started

Anyone starting without experience has two clear candidates — and one that you shouldn't start with, as tempting as it sounds.

Typical entry point 1 — mail routing

Immediately tangible relief in the office, low risk, quick to measure. Once that runs, the team has the confidence for the next steps.

Typical entry point 2 — quote template

Biggest lever for the owner, because evening work and lost jobs pile up here. Needs more preparation (cleanly structuring standard clauses and hourly rates once), but pays off on every quote.

Don't start with the maintenance contract

As tempting as well-kept contract dates sound, the first step often fails on messy master data. Tidy up first, then automate.

Funding for trades businesses

funding for SMBs is regionally tiered: 50% or 80% of consulting costs — in our service area Hameln/Hannover/Hildesheim/Bielefeld it's 50% (max. 1,750 € subsidy), in Lüneburg region 80% (max. 2,800 €). Trades businesses are a main target group and applications are usually decided within a few weeks. Chambers (HWK) also offer their own digitalisation programmes — but note the cumulation prohibition with other public subsidies. For larger implementation investments, ZIM or go-digital are separate funding tracks.

→ Details on BAFA funding
FAQ

Frequently asked questions about AI in the trades

No. We build solutions used through familiar paths — smartphone photos in WhatsApp, incoming email, voice notes. Your fitters and office staff see no interface and keep working with the tools they know.
Yes, as soon as repetitive office work exists: quotes from on-site measurements, invoice reconciliation, call notes, time tracking. Typical saving: 5 to 12 office hours per week. With funding it usually amortizes in the first quarter.
A fitter takes 3 to 5 photos on site with the smartphone and speaks two sentences. The recognizes the room, materials and dimensions and creates a draft quote in your CRM. You review, adjust prices, send. 5 minutes instead of 30 to 45.
For end-customer data we use or EU-hosted models (GDPR Art. 28 DPA, no transfer to the US). For B2B customers cloud is usually unproblematic — we clarify this in the initial call and document the choice in the model example.
We work with what is already there. Most craft software has CSV export or an . If not, we go via email workflows or browser automation. does not replace your management software, it feeds it faster.

Concrete model example

How an electrical business with ten workstations builds its AI mix

Three roles — management, site lead, back office — get different tools with different data sensitivities. Frontier cloud for strategy and mobile, for back office and knowledge base. Documented as a model example that transfers to other trades.

→ The model example in detail

Ready for the next step?

Free intro call, no strings attached. In 30 minutes you'll know whether and how AI can help your business.

Book a callCheck eligibility
Book a callBAFA funding