Before You Quote the Job: How New Techs Can Use AI to Get Sizing Right

Somewhere out there, a contractor with a three year old business just told a homeowner that her ductwork was fine because the new system was the same tonnage as the old one. He might be right. But he never showed his math, and now dozens of technicians in an online group are picking apart an answer he gave in thirty seconds. That is not a story about one bad contractor. It is a preview of what happens to anyone who quotes a sizing job on a rule of thumb instead of a calculation, and if you are early in your career, it is exactly the mistake worth avoiding before it has your name on it.

You do not need ten years in the field to avoid it. You need a habit of running the numbers before you speak with confidence, and AI can genuinely help with that, as a second set of eyes, not as the technician doing the work.

What AI Is Actually Useful For Here

Used correctly, an AI assistant is good at exactly the part of sizing work that is tedious and easy to get slightly wrong under time pressure: organizing your inputs, running the standard conversions, and holding a set of numbers steady while you check them against each other.

Organizing your inputs into a clean estimate. Feed it square footage, ceiling height, insulation era, window count and orientation, and climate zone, and it can walk you through a rough load estimate to sanity check against, before you're standing in front of a customer doing arithmetic in your head.
Running the standard conversions without a mistake. Tons to CFM, CFM split across rooms by relative load, duct diameter against target airflow. Simple math, easy to fumble when you're rushing between calls.
Flagging a mismatch you might not think to check. Tell it the new equipment is variable-speed or inverter-driven, and a good back-and-forth will surface that static pressure sensitivity is a real consideration, the same point that tripped up the contractor in that Facebook thread.

What AI Is Not Good For

⚠️ It has never been inside the house. It does not know if your duct insulation is falling apart, if a branch run has three unnecessary 90 degree turns, or if the attic your air handler is going into hits 140 degrees in July. It can sound completely confident while working from bad or incomplete information, and it has no way to tell you when that is happening. That is your job, not its job.

An AI assistant is not a substitute for actual Manual J and Manual D software, it is not a substitute for a static pressure reading with a manometer, and it is not a substitute for a senior tech looking over your numbers before they go on a proposal. Treat its output as a draft to verify, never as the answer you hand to a customer.

A Practical Way to Use It

The value comes from what you feed it. Vague questions get vague answers. Specific inputs get a specific, checkable estimate.

Example, Sizing Sanity Check

Give It

Square footage and layout, ceiling height, insulation type and approximate age, number and orientation of windows, climate zone or city, existing duct sizes and approximate run lengths, and whether the new equipment is single-speed or variable-capacity.

Ask For

A rough total CFM requirement, a suggested room-by-room airflow split, and whether the existing duct sizes look adequate to carry that split, with the reasoning shown, not just a conclusion.

Then Verify

Run the same job through an actual load calculation tool, and confirm the duct sizes and run lengths in person rather than trusting what you told the AI they were from memory.

Always Verify Before You Write the Number on a Quote

Before anything from this process ends up on a proposal, it needs to survive contact with a real calculation, not just a chat window. That is exactly why we built our own Manual J load calculator, use it as the check on your work, not as your first draft, and definitely not as your only draft. If you are new enough that you are not fully confident reading the output yet, get a senior tech to look at it with you. That is not a sign you are behind. It is exactly how everyone above you learned to do this correctly.

The tech who runs the numbers before quoting the job never has to explain his math to a stranger after the fact. Build that habit now, while it's still cheap to build.