Use cases by industry4 use cases

AI automation for contractors, subcontractors and trades

A 90-person electrical subcontractor with 25 live projects loses money in two quiet ways. Changes buried in daily logs, emails and drawing revisions get built and never priced, and when a claim is finally raised, the contractual notice deadline under VOB/B, JCT, NEC or an AIA contract has already passed. A change-order notice guard watches for likely scope changes and computes notice deadlines from the contract, and the project manager sends every notice.

General contractors on Procore already have native agents for RFIs, submittals, daily logs and contract review, and Trimble bought Document Crunch in April 2026; if you are on Procore, look at those first. Much of the European mid-market is not. German trades run HERO, Plancraft, Craftnote or openHandwerk, price GAEB bills of quantities, and get few AI features from their software.

In the trades, paperwork rather than demand is often the bottleneck. A German solar and heat pump installer files grid-connection requests with hundreds of grid operators, each with its own portal, registers systems in the Marktstammdatenregister and handles KfW subsidy files; installer grid paperwork fills those forms and stops before submission. The phone is the other leak, and an after-hours call agent books jobs and puts a gas smell or a burst pipe straight through to the on-call technician.

What I would build for construction and trades

Where builders and installers leave money on the table

  • Bids nobody had time to price

    Estimators read invitations, specs and drawings by hand, so capacity caps how many tenders get priced. Go or no-go scoring against past wins, with every requirement pulled into a matrix that cites its page, as the public tender assistant does, decides where the estimator's time goes.

  • GAEB bills of quantities

    Takeoff tools such as Togal and Beam AI measure drawings well. Pricing a GAEB bill of quantities from your own historical costs and supplier quotes, with exclusions flagged, is company-specific logic they leave to you.

  • Subcontractor paperwork before payment

    Insurance certificates, lien waivers and pay applications are checked against the contract before money goes out, so an expired certificate is caught at invoice time rather than after an accident.

  • Daily reports from the site

    A foreman's voice note and photos become a structured daily log with weather, crews and delays, which is also the evidence a change-order claim will later depend on.

What the site team and the office keep for themselves

  • Sending notices and claims

    The agent drafts the notice with photos, log excerpts and the drawing difference. Contract notices have formal requirements, often written and to a named person, so the project manager checks and sends each one.

  • Prices on the phone

    An after-hours agent books holds within your rules. It never quotes a price or promises an arrival time the schedule does not show, and it says at the start of the call that it is an AI.

  • Who gets woken up at night

    The on-call rota in code, not the model, decides who is called at 2am. A gas smell gets the gas emergency number and safety instructions first, a burst pipe gets a warm transfer, and a dripping tap gets a slot on Tuesday.

  • Logins and CAPTCHAs on grid portals

    Browser automation fills operator portals but leaves logins, MFA and CAPTCHAs to a person, logs a screenshot at each step, and raises an alert when a portal changes instead of guessing.

Software these builds usually connect to

The systems do not get replaced. The build sits across them, reads from them through their APIs, and writes results back.

Procore / Autodesk Construction Cloud / Bluebeam / Raken / Fieldwire / RIB iTWO / Nevaris / HERO / Plancraft / Craftnote / ServiceTitan / OpenSolar

Frequently asked questions

Can AI track change orders and notice deadlines on construction projects?

It can watch daily logs, emails, RFI responses and drawing revisions for likely scope changes and delays, attach the evidence, and compute notice deadlines from contract terms that were extracted once and confirmed by the project manager. It should not send notices: they carry formal requirements and legal weight, so the draft goes to the project manager, who decides and sends.

Is there an AI receptionist for plumbers and heating engineers?

Several, and for simple booking they are enough: Smith.ai, Rosie and Goodcall in the US, and ServiceTitan sells an AI voice agent as a Pro product. Trades need more than booking, though: emergency triage for gas and water, dispatch rules, European trade software and callers with strong dialects. That is where a custom agent on your own schedule and routing rules makes sense.

Can AI fill in grid-connection applications for solar installers?

Yes, with a person in the loop. An agent pulls job data from the planning tool, fills each grid operator's form or portal, stops before submission for review, tracks status and chases missing items, then prepares the Marktstammdatenregister entry. Start with the five operators you file with most, because every operator's portal is its own template and changes without notice.

We use Procore. Do we still need custom AI?

Maybe not. Procore's own agents cover RFIs, submittals, daily logs and contract review, and they already know your Procore data. Custom work fits around the edges: subcontractors who work in their customers' Procore but run their own business elsewhere, notice logic under VOB/B, GAEB pricing, and the office side, from tenders to the phones.

Also browse: Operations and logistics · Sales and lead handling · Outlook and Microsoft 365 · Real estate and property · all use cases