Daily field briefing
POLR AIPOLR AI

AI + AEC Today

Daily AI intelligence for the built world

Weekly digest

Weekly AI in AEC stories

← Archive

Edition · Thursday, July 23, 2026

AI & Construction

Weekly Pulse · Industry actionability

The field is piloting, not just experimenting.

62%
ExperimentalPilotingField-ready
  • Firms are being told to rebuild their work around AI, not just buy apps
  • AI is moving into BIM to catch clashes before crews build
  • AI is starting to coach and train workers, not just answer questions
  • The AI data center boom keeps feeding construction work

Practical Use of the Day

Find the busywork in one job that AI could take over

AI helps most when you rebuild a whole task around it, not when you bolt it on. Pick one job your team does the same way each week, like logging RFIs or checking invoices. Write out each step. Then ask AI which steps are rote and rule-based, which are a good fit for AI, and which still need a person. Start with the rote steps. That is the fast, safe way in.

Try this prompt

You are helping a construction team find where AI can save time. I will describe one routine task we do, step by step. First, list each step and mark it as rote and rule-based, or as needing human judgment. Second, tell me which steps are the best fit for AI to do or draft, and why. Third, tell me which steps a person must still own so we do not add risk. Keep it plain and short. Here is our task:

Tool — Claude (or any capable LLM)

Stories

4 curated
ConTechHigh impact

McKinsey: the AI winners will redo how work gets done

McKinsey put out a new report on AI in building. It says AI will not wipe out firms, but it will change how work gets done. The firms that win will redo whole tasks around AI, not just add a tool here and there. McKinsey found AI could do about 39% of the desk work in construction. It says to build your own tools only where your skill is the real edge, and buy the rest.

Why it matters to GCs

This is a plan, not just hype. It tells you where to start: bids, estimates, and proposals in the next year or so. It also warns that just buying AI apps will not help much if the work around them stays the same.

Construction Dive·Jul 22·Read source →

LinkedIn
ConTechModerate

New AI service turns flat plans into build-ready 3D models

Attentive.ai launched a service called BIM CoPilot. You send it flat PDF or CAD plans. It builds a coordinated 3D model and a clean drawing set your field crew can build from. AI helps find where trades clash, like a duct running through a beam. A human BIM team checks every step before it comes back to you.

Why it matters to GCs

A clash found on paper costs far less than one found in the field. Catching them before work starts cuts change orders and rework. The human check means you are not trusting the whole model to a bot alone.

Geo Week News·Jul 1·Read source →

LinkedIn
AI FrontierModerate

AI avatars now coach workers through tough talks

Synthesia makes AI video avatars for training. Now it added live practice sessions. A worker can rehearse a hard talk, like a safety warning or a bid review, with an AI avatar that talks back. The avatar pushes back, then scores how the worker did. Bosses get a report on how the whole team is doing.

Why it matters to GCs

Construction is short on skilled people, and real training takes time. Letting crews practice with AI, on their own schedule, can build soft skills faster. Just remember the AI scores talk, not real field work.

TechCrunch·Jul 22·Read source →

LinkedIn
AI & BusinessModerate

OpenAI to spend $750B on AI, feeding a data center boom

OpenAI, the maker of ChatGPT, said it will spend $750 billion on AI through 2030. Most of that goes to giant data centers full of computer chips. Its next site in Georgia covers 1,400 acres and needs as much power as a small city. That is a huge amount of new building work.

Why it matters to GCs

Data centers are one of the hottest jobs in construction right now. Deals this big mean years of work for firms that can build them and the power to run them. It is also a sign that AI spending is not slowing down.

TechCrunch·Jul 22·Read source →

LinkedIn