Here We Go
This month we are taking our heads out of the clouds and getting PRACTICAL.
I typically incorporate a mix of practical tools, industry trends, and my own experiences and experiments - but I thought it would be useful for my fellow builders to see some very basic examples of how to start actually using AI.
The reality I’m confronted with in nearly all my client conversations, is that we are still in the early adopter phase. Most folks I talk to have tried ChatGPT and may even use it regularly for things like email drafting, but that’s often where it stops.
The goal of this practical edition is to help people stuck in level 0-1 of AI usage start experimenting more deeply. Experience has shown me that once you start experimenting a little bit, the use cases pile up fast.
So here we go - 5 practical AI experiments you can run today on your free account with any major provider:
EXPERIMENT #1
Turn Any Project Doc Into Your Personal Q&A Assistant
Why it works: You've got PDFs everywhere. Contracts, product install guides, code books, plan notes. Finding one detail means opening the file, scrolling around, maybe Ctrl+F and get lucky. Instead, upload it once and just ask questions.
Try this in ChatGPT, Claude, or Gemini:
Upload any project doc or PDF you're working with
Ask it questions like:
"What's the required R-value for the exterior walls on this project?"
"What does the contract say about allowances for lighting fixtures?"
"What are the flashing requirements at the window-to-wall transitions?"
Pro tip: After uploading, ask: "Summarize the 10 most important things I need to know from this document."
Success metric: "Did I find the answer faster than opening the file and hunting for it?"
EXPERIMENT #2
Compare Two Sub Bids Side By Side
Why it works: When you get two bids for the same scope, the differences hide in the details. One sub includes dumpsters, the other doesn't. One excludes winter conditions, the other buries it in fine print. AI can spot what you might miss on a quick read.
Copy and paste this into ChatGPT, Claude, or Gemini:
Paste or upload both bids into the same conversation
Ask it questions like:
"Compare these two bids. What does one include that the other doesn't?"
"Which bid has more exclusions?"
"Create a table showing the differences in scope, pricing, and exclusions"
"What questions should I ask each sub before making a decision?"
Pro tip: After the comparison, ask: "If I go with the lower bid, what risks am I taking based on their exclusions?"
Success metric: "Did I catch something I might have missed otherwise?"
EXPERIMENT #3
Generate Tomorrow’s Safety Meeting
Why it works: Safety meetings are required but often rushed or skipped. Having one ready the night before means you might actually do it.
Copy and paste this into ChatGPT, Claude, or Gemini:
Create a 10-minute safety meeting for a construction crew.
Topic: [excavation safety / ladder safety / heat stress / etc.]
Work plan for tomorrow: [e.g., "pouring footings" or "framing second floor"]
Format:
- 3 key safety points
- 1 real example or scenario
- 1 question to ask the crew
- Keep it under 200 words
Try it right now with: "Topic: Fall protection. Work plan for tomorrow: roof sheathing."
Success metric: "Did we actually do the safety meeting instead of skipping it?"
EXPERIMENT #4
Turn Voice Notes Into A Daily Log
Why it works: Every PM knows the feeling. It’s the end of the day, you're tired, and you still have to write a daily log. Most people are already talking into their phone on the drive home. Now you can turn that ramble into a clean report. This one has change my own workflow when running projects.
Copy and paste this into ChatGPT, Claude, or Gemini:
Turn these rough voice notes into a clean daily construction report.
Format:
- Date and project name
- Weather conditions
- Crew count by trade
- Work completed today
- Issues or delays
- Plan for tomorrow
Voice notes:
[paste your transcribed voice notes here or simply click the microphone icon and start talking.]
Try it with these sample voice notes:
"Yeah so today was good, had 4 framers and 2 plumbers on site. Weather was decent, mid 40s, rained a little after lunch. Got the second floor walls stood up on the east side. Plumbers started roughing in the first floor bathrooms. Had to wait on the beam delivery til 10 which pushed framing back a bit. Tomorrow we should finish east side walls and start south if the beam situation is sorted out. Oh and the inspector is coming Thursday for the footing inspection on building B."
Success metric: "Did I actually submit a daily log today instead of putting it off?"
EXPERIMENT #5
Sanity-Check An Estimate
Why it works: We’ve all missed a line item on an estimate. AI won't give you exact numbers for your market, but it's surprisingly good at spotting what's missing and flagging things that look off.
Copy and paste this into ChatGPT, Claude, or Gemini:
Review this rough estimate for a construction project. Tell me:
1. What line items seem to be missing?
2. Does anything look unusually high or low?
3. What would you expect to see for this scope that isn't listed?
4. Any general red flags?
Project description:
[brief description of the work]
Estimate:
[paste your estimate here]
Try it with this sample:
"Bathroom remodel, gut to studs. Demo: $1,500. Framing: $2,000. Plumbing rough-in: $3,200. Electrical: $1,800. Drywall: $2,100. Tile (floor and shower): $4,500. Vanity and fixtures: $2,000. Paint: $800. Total: $17,900."
Pro tip: After the review, ask: "What's a realistic contingency percentage for this type of project?"
Success metric: "Did AI catch something I forgot before I sent this to the client?"
I want to hear from you. If you try one of these, send me an email and tell me which one you picked and what happened.
And if one of these experiments sparks something bigger and you want help building it into your daily workflow, please reach out.
Murray
P.S. Know someone in construction who'd find this useful? Give it a share.
