Ongoing experiment
This is not a feature list or a pitch for a specific tool. It is a practical look at where AI is proving useful in mechanical construction — and where the output still needs experience, judgment, and verification.
Where I’m using it
A closer look at how I’m using AI to turn dense project information into cleaner checklists, faster drafts, better first-pass reviews, and more efficient day-to-day communication.
01
AI-Assisted Spec Checklists
Using AI to review drawings and specifications, I can generate project-specific checklists that summarize key material requirements, system requirements, pipe size breakpoints, and equipment requirements. It took some trial and error to develop a prompt that produces a usable result, but once dialed in, it can turn dense specs into a clear working checklist.
One of the most useful parts is that it can help surface obscure requirements that may be buried deep in the specifications or tucked away in a small drawing note. Each item includes the reference location where the information was found, so I can verify it directly against the contract documents myself. That makes the checklist faster to build, easier to use, and still grounded in the actual project documents.
02
AI-Assisted Submittal Review
AI is helpful for a first-pass review of submittals against the project specifications. It can quickly create a list of possible discrepancies, missing information, or items that need a closer look.
The final review still requires experience and judgment, but AI helps speed up the cross-referencing process and saves time.
03
AI-Assisted RFI Drafting
AI is useful for quickly creating a first draft of an RFI. I can talk through the issue by voice much faster than I can type it, then use AI to organize the information into a clear, professional draft with the right context, question, and supporting details. The framing still needs human judgment, but the time to a clean draft is dramatically shorter.
04
AI-Assisted Email Drafting
One of the biggest time savers is AI-assisted email drafting. Mechanical project management requires a constant flow of emails, and AI can turn rough notes, voice input, or quick bullet points into a clear, professional draft much faster than starting from scratch.
The key is setting up the right custom instructions so the language matches the way you actually communicate. Once that is dialed in, it becomes a major time saver for day-to-day project communication.
I’m not trying to automate judgment. The real value is using AI to reduce repetitive cognitive work, so more time and attention can go toward the decisions that require experience, context, and good project judgment.
Where it still falls short
There are many areas where AI is developing quickly, and it will be capable of estimating projects very soon. But in my experience, it is not reliable yet for quantity takeoffs from drawings. It also hallucinates on technical specs with enough confidence to be dangerous if you’re not checking the output. I check everything.
This page will keep getting updated as the tools improve and my use of them does too.