Brief sections
The leak
An RFQ arrives Monday morning. The one estimator who knows that product line is on site until Thursday, so the request waits. The quote goes out on day six — thorough, accurate, and too late, because the customer signed with whoever answered on day two. Meanwhile the pricing logic itself lives in one person’s head and a spreadsheet with seventeen tabs, which means two estimators can price the same job meaningfully apart and both defend the number. The cost shape is triple: bids lost to speed, margin lost to inconsistency, and a key-person risk that gets priced into your valuation if a buyer ever looks.
What AI realistically does here
Not an AI that prices work unattended. A drafting system: it reads the request — plans, spec sheets, photos, the email thread — pulls comparable jobs from your history and the current catalog and price list, and assembles a draft estimate with every assumption listed for review. The estimator’s judgment stays in the loop; the hours of lookup and assembly leave it. Where pricing rules are genuinely rules — waste factors, minimums, freight bands — they get encoded and versioned instead of remembered.
What has to be true first
- The pricing logic can be written down. If two estimators price the same job 20% apart and both are “right,” that’s a pricing-policy decision for the owner, not an AI project — we’d settle it in the workflow redesign before any software gets built.
- Past quotes are findable and connected to outcomes, even loosely. Won/lost plus final price is enough to start; a folder of PDFs beats nothing.
- A reviewer with authority. Every draft estimate crosses a human desk until the error rate has earned trust, and high-stakes bids cross one forever.
What we'd install
The estimate copilot doesn’t ship alone. The opportunity gets scored first — value, data readiness, adoption path — against everything else leaking in the business, because quoting is often but not always the top leak. Then the pricing rules move out of heads into a versioned SOP, the copilot goes on the roster with a named owner (usually the senior estimator), and quote turnaround and hit rate get lines in the value ledger, reviewed with leadership. The full sequence is the AI operating partnership.
Worth it?
Pull your last twenty bids and measure the days from “request received” to “quote delivered,” plus what an estimator-hour actually costs you loaded. The calculator below turns hours and rates into an illustrative annual figure — illustrative math, not a promise. Whether your quote history and pricing data can support the build is what the scorecard checks.