The Beast from the East arrived in February 2018 and the Federation of Master Builders reported that 58% of UK builders had to stop work entirely. Billions of pounds of productivity evaporated in a week. Nobody was surprised by the weather. They just had no tool that told them, at site level, how bad it would get and when.
That is the gap AI weather forecasting is trying to close. And the results, for the teams that have adopted it seriously, are worth paying attention to.
What standard forecasts miss
A BBC weather forecast is good enough for deciding whether to take an umbrella. It is not good enough for deciding whether to pour concrete, operate a tower crane, or schedule a roofing team.
Weather conditions vary significantly over short distances. A site in a river valley behaves differently to a site 2 miles away on exposed high ground. Wind patterns around a high-rise under construction bear no resemblance to the regional forecast. Moisture conditions for concrete curing depend on temperature, wind speed, and humidity working in combination, not any single variable.
Regional met forecasts are built around population centres and averaged over wide areas. Construction projects need hyper-local precision. That is what AI platforms are now delivering.
The tools doing the work
Tomorrow.io's weather intelligence platform uses machine learning combined with satellite data to produce hyper-local near-real-time forecasts. It provides automatic alerts for hazardous conditions, integrates historical weather data for pre-project planning, and allows project managers to schedule weather-sensitive tasks around predicted windows of favourable conditions. Crane operations, scaffold installation, concrete pours, and external cladding work all have specific weather tolerances. Tomorrow.io lets teams build those tolerances into their scheduling.
Foresight, a London-based predictive delivery platform that integrates with Primavera P6 and Microsoft Project, goes further. Its analytics include weather pattern analysis as part of broader delay risk prediction, and it has demonstrated 2x more accurate delay forecasting and 30% overrun reductions with clients including Google and Compass Datacentres. The system flags weather-driven schedule risks weeks before they materialise, rather than on the day the rain arrives.
The Weather Delay Problem in UK Construction
95%: of UK construction projects experience delays in 2025 (Elecosoft)
58%: of UK builders paused work during the Beast from the East (Federation of Master Builders)
20-30%: cost overruns on delayed UK construction projects
Safety stand-down decisions
There is a harder question underneath the scheduling conversation. When do you stop work for safety?
Crane operations in the UK typically stop at wind speeds above 38mph at hook level. But wind speed at ground level and wind speed at 60 metres can differ substantially, and standard forecasts do not give you hook-height data. AI weather tools that integrate IoT sensors with forecasting models can. They can alert a site manager that conditions at crane height are approaching threshold two hours before it becomes critical, rather than after the incident.
The same logic applies to scaffold stability, roofing operations, and any work with temperature or moisture requirements. Climate-related events cost the UK construction industry billions annually. The 2015-16 winter floods alone caused an estimated £1.6 billion in construction damages. Proactive weather intelligence does not eliminate weather risk. But it turns a reactive crisis into a planned stand-down, which costs a fraction as much and carries no safety risk.
"A proactive stand-down planned 48 hours ahead costs a fraction of an emergency halt. AI forecasting makes that the default."
Getting started without the overhaul
You do not need to rebuild your project management system to get value from weather AI. Most tools integrate with existing scheduling software. The starting point is picking your highest-weather-sensitive projects, typically those with significant outdoor concrete, cladding, or groundworks programmes, and running a pilot that compares AI forecast-driven scheduling against your existing approach.
Measure two things. Unplanned weather shutdowns versus planned ones. And the percentage of weather-sensitive tasks completed in the originally scheduled window. Those two numbers will tell you whether the tool is earning its place.
Elecosoft, whose Asta Powerproject is used on major UK infrastructure schemes, notes that integrating real-time data with scheduling gives planners the ability to model weather-driven scenarios and re-sequence work hours, not days, after conditions change. On HS2 and similar large-scale projects, that kind of responsiveness is already standard. On a £5 million commercial fit-out, it still is not. But it could be.
Building AI Competence in the Built Environment
The Responsible with AI programme equips built environment professionals to evaluate, procure, and govern AI tools. Explore how to build the right frameworks for your firm.
Explore the Programme → Responsible with AI
