FlowsBook a callSign inCreate free accountSign up
AI employees and automation

An AI agent for construction companies: new projects, closures and the forecast

How an AI agent for construction companies finds new projects to bid on, warns of road closures near active sites, and reads the forecast before pours and crane lifts.

Part of the guide AI employees

By the xMap team · · 5 min read

An AI agent for construction companies does three jobs well: it collects new projects and building activity your estimators should know about, it warns site managers about road closures and roadworks near each active site before deliveries leave, and it reads the weather forecast for every site so pours, crane lifts and outdoor work are planned around rain and wind. In xMap Flows each job is an AI employee that runs on a timetable and lands in the inbox of the person who acts on it.

It is aimed at contractors, subcontractors, fit-out firms and building product suppliers, from a regional builder with four sites to a group with forty. Where a block only works in some places, that is said in the section.

New projects worth bidding on

Work rarely appears in one place. Some of it is in trade news, some on the public pages of planning departments and procurement portals, and some shows up first as a company hiring site engineers. Three blocks gather it:

  • News reads trade news feeds or a news page and, with a few words such as tower, warehouse or hospital, keeps only the stories about projects. It can also ask xMap AI to keep only news about the kind of work you do.
  • Web page watch reads up to 10 public pages each run, such as a planning register or a council's list of approved schemes, and passes on only the new entries since last time.
  • Job postings finds other contractors hiring site engineers or quantity surveyors in your region, a sign they have won work you may be able to subcontract on.

An xMap LLM block then writes a short note on each lead: what it is, where, and whether it fits your size of job. The contractors winning work template uses the hiring signal, and new villa and apartment projects for fit-out work suits finishing trades. In some United States counties, the Property changes block adds homes that just sold, a steady source of renovation leads.

Road closures near each active site

A concrete truck stuck behind a closure that nobody checked is an expensive morning. Put your active sites in a table in My data with their addresses. The Traffic incidents block, set to check around each place that comes in, gives every site the number of closures, roadworks and incidents within the distance you choose, and the nearest one with its street.

Run it before the first deliveries, for example at 05:30, and send only the sites that have something nearby. A Filter on the number of incidents does that, so a quiet morning means no email at all.

The forecast, site by site

The Weather forecast block gives each site the coming days' rain, temperature and the strongest wind, up to 16 days ahead, plus the next day it expects rain. A Filter can then keep only the sites where the forecast crosses your own limits: rain on a pour day, or wind above the speed your crane supervisor works to.

The morning site briefing template puts closures and weather together in one email per morning. For crane work, wind alert for crane lifts checks the wind every hour, and rain ahead: plan the concrete pours looks at the next few days.

Example: one contractor's week (fictional)

Take a made-up contractor with five active sites around Nairobi: two apartment blocks, a warehouse, a school extension and a road junction upgrade.

On Monday the briefing email lists two sites. The warehouse site has a road closure 400 metres away on the main access road, so the site manager moves the steel delivery to the afternoon. The school extension has 18 mm of rain forecast for Wednesday, so the slab pour moves to Thursday.

On Tuesday, the leads email holds three items: a news story about a new hospital wing in a nearby town, a new entry on a county's list of approved schemes, and a mid-sized contractor hiring two site engineers. The estimating lead calls the contractor about subcontract work on its new job.

Where teams go wrong

  • Checking closures across the whole city. A list of every roadwork in a capital is noise. Check around each site, within the distance a delivery truck actually uses.
  • Running the forecast once a week. Forecasts change. A morning run for the next few days, plus an hourly wind check on crane days, is what helps.
  • Leads with no filter for size. A small builder does not need every stadium story. Tell the LLM your typical job size and the kinds of work you take on.
  • Forgetting finished sites. Keep the My data table current, or the briefing keeps reporting closures near a site you left last month.

Setting up the AI employees

  1. Site briefing: Schedule (weekdays at 05:30), then My data (active sites), then Traffic incidents (around each place, within 1 km), then Weather forecast (each place, the next 3 days), then Filter (only sites with an incident or rain above your limit), then Email to the site managers.
  2. Crane wind: Schedule (every hour on lift days), then My data (sites with a crane), then Weather forecast, then Filter (wind above your limit), then Email to the crane supervisor.
  3. Project leads: Schedule (weekdays), then News and Web page watch (the feeds and public pages you follow) and Job postings (site engineers in your region), then xMap LLM to judge fit, then Only new, then Email to the estimating lead.
  4. Run each once, read the email it would send, then Deploy. All three appear in the Team inbox, where you can ask each one about what it found.

If your teams spend their day on sites and on the road, the guide to AI employees for field operations covers more jobs like these.

Questions

Does it know about closures that are planned for next month?

Traffic incidents shows what online maps report right now: closures, roadworks and other incidents. Planned works that are only announced elsewhere can be followed with Web page watch on the authority's public page.

Can it alert on wind at a particular height?

The Weather forecast block gives the forecast wind near the ground for each site. Use it as an early warning, and keep your own anemometer readings for the lift decision.

Will it work for sites outside cities?

Yes. Closures and the forecast work anywhere with a position. Online maps report fewer incidents on rural roads, so treat a quiet result there with care.

Where do the leads come from?

From news feeds and public pages you choose, and from public job postings. Property changes adds recent sales in the US counties it covers.

Templates that do this

Read next

More on ai employees and automation · Every article · Templates

Put it to work

An AI employee in xMap Flows does this on live data, on its own timetable. Start from a template or describe it in plain words.