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, September 24, 2026

AI & Construction

Weekly Pulse · Industry actionability

The field is piloting, not just experimenting.

58%
ExperimentalPilotingField-ready
  • AI agents now draft bids, daily logs, and drawing checks for building teams
  • New AI models do longer, harder jobs for less money
  • AI starts to do real office work, not just answer questions

Practical Use of the Day

Compare bids and catch scope gaps with AI

You do not need special software to line up bids. A plain AI chat works. Take two or three bids for the same scope. Paste them into the chat. Ask the AI to put them side by side. Ask it to list what each one includes and leaves out. Then ask where the scopes do not match. The AI can flag a gap, like one bid that skips the cleanup. Now you can ask the right questions before you pick.

Try this prompt

I am a general contractor comparing subcontractor bids for the same scope of work. Here are the bids: [paste each bid]. Put them side by side in a table. For each bid, list what is included, what is excluded, the unit prices, and the total. Then tell me where the scopes do not match, and flag any gaps or items only one bid covers. Point out anything I should question before I award the work. Use plain language.

Tool — Claude (or any capable LLM)

Stories

3 curated
ConTechHigh impact

Autodesk shows AI agents that draft bids and daily logs

At its big user show, Autodesk showed new AI helpers for building work. One helper, called Forma Bid, lets you compare bids and spot gaps in scope. Another turns a super's spoken notes into a daily log. A third checks drawings for problems, like walls set too close together. People still review the work and make the final call. A wide launch is planned for 2027.

Why it matters to GCs

Bids, logs, and drawing checks eat hours every week. AI that drafts them first hands that time back to your team. It can also catch scope gaps and drawing errors before they cost you.

Architosh·Sep 16·Read source →

LinkedIn
AI FrontierModerate

Anthropic's new AI works on hard jobs for hours on its own

Anthropic put out a new AI model called Claude Opus 5.5. It is built to work on long, hard jobs with little help. In one test it finished a huge coding job in under a day. It also costs about 20% less to run than the last model. On a full task it can cost 40% less. It is also harder to fool with sneaky messages.

Why it matters to GCs

The AI inside your tools keeps getting stronger and cheaper. So the helpers you buy can do more work for less money. Better guards also make these tools safer to trust with job data.

VentureBeat·Sep 22·Read source →

LinkedIn
AI & BusinessModerate

AI workers get paid only when they finish the job

A startup called Ema raised $77 million for its AI workers. These AI agents do office jobs in HR, IT, and finance. Ema does not charge by the seat or by use. It charges only when the AI finishes a task. Big firms like PwC and Wipro already use it. Ema says more than a million people now use its AI workers.

Why it matters to GCs

Your back office runs on software you pay for by the seat. AI that charges by the job could change what that software costs. It also shows AI can take on real paperwork, not just chat.

TechCrunch·Sep 23·Read source →

LinkedIn