Use AI to organize commercial HVAC change-order requests

Keep the approved baseline intact

Start with the accepted scope version and any previously approved changes. Collect the new request, its source, date, requestor and affected location or equipment. Preserve the original message or field record rather than replacing it with an AI paraphrase. A change log should make it possible to reconstruct what the team knew at the time.

Ask AI to separate newly requested work, clarification of existing scope and uncertain differences. Do not let it decide whether a contract requires additional payment. Keep the estimator, project manager and appropriate commercial reviewer responsible for that decision.

Build a review packet before a price

Use fields for baseline reference, requested difference, supporting evidence, affected work, unanswered technical questions, potential scheduling dependencies and review owner. Affected work should describe what may need examination, not a generated instruction to the field. Preserve customer wording where the exact request matters.

ServiceTitan documents a change-order workflow for supported project contexts. That is background for record handling; this proposed AI intake does not certify entitlement, calculate a price, change a project budget or bypass permissions. Verify account configuration before using any named product workflow.

Treat a late request as a question, not a commitment

For a fictional tenant fit-out, a customer asks to add a second control zone after the proposal was accepted. The intake packet should point to the approved zone count, record the new request and assign technical and estimating review. It should not choose components, imply the original design was deficient, or promise the existing completion date will hold.

If the request turns out to be within the original scope, record that reviewed outcome and its reasoning. If a change is proposed, keep the proposed document distinct from the customer-approved change. Do not advance its status because someone said it looked reasonable in a meeting.

Use a small retrospective to test extraction

Review a handful of closed change requests with a project manager. Check whether AI found the correct baseline, preserved request dates, identified duplicates and avoided inventing cost or delay. Record false change flags as well as missed changes. Measure review time and the number of questions that needed another document search.

Keep final approval, pricing and customer communications within the established process. When a change is approved, link it back to the original intake record and revise the operational handoff explicitly. Never rewrite the accepted baseline as though the changed scope had always been there.

Change-request checklist

• Accepted baseline and prior approved changes are identified.
• Requestor, date, location and original evidence are retained.
• New work, clarification and uncertain differences are separate.
• Technical, estimating and schedule reviewers are assigned.
• Only authorized approval changes the scope or customer commitment.