All plays › Local trades and home services

Local trades and home servicesClaude

Build a Claude ballpark-estimate tool from your past jobs

Picked and checked by Olga Pearson, host of The Practical AI Show · Published

Result

as reported

An estimate that took 8 hours to put together took his AI tool about 30 minutes, and came in within 5% of the estimate his staff prepared.

Mathew Meskimen owns MCG Homes, a remodeler in Fernandina Beach, Florida (about $6 million in 2025 revenue) and a 2026 Fred Case Award finalist. MCG picked Claude as its main business AI and built shared projects for estimating, sales qualification, HR, SOP search and construction standards, so no one has to rewrite prompts. His ballpark tool lets a salesperson describe a scope and property, compares it with MCG’s past jobs and market data, and drafts a budget with MCG’s cost codes, allowances and contingencies.

Who did it

Mathew MeskimenMCG Homes (remodeler, Fernandina Beach, Florida)Qualified Remodeler · Oct 8, 2026

Try it now

  1. Export 20 or more finished jobs with scope, square footage, final cost and your cost codes.
  2. Make a shared Claude project with that data, your cost codes, your allowance rules and instructions to ask for missing details.
  3. Have a salesperson describe a new scope and property, and ask for a ballpark range plus a draft budget by cost code.
  4. Test it against one estimate your team already built the normal way before anyone relies on it.
  5. Use the range to talk budget honestly with the client before design starts.
Copy-paste prompt
You are the ballpark estimator for [company]. Use only the past jobs, cost codes and allowance rules in this project. New lead: [scope], [property details], [client budget]. Give me: (1) a low / likely / high range with the 3 most similar past jobs; (2) a draft budget by our cost codes with labor, subs, materials, allowances and a contingency that fits the risk; (3) whether the client’s budget is realistic and where they’d need more money. If key details are missing, ask before estimating.

Make it run every week

A play works once. Turn it into a routine in three small steps:

  1. Run it once and check the result yourself.
  2. Give your AI a schedule. Tell it when to run this play (every week, or every month) and to message you with what it found before you act.
  3. Keep a log. One line per run: what was checked, anything flagged, what you did.

See how owners turn plays into systems →

What to watch out for

Reported by a trade magazine from his award application. One side-by-side test, within 5%. Keep checking it against real estimates, since the tool is only as good as your past job data. Build it with Page: a “what does a remodel cost here?” page that books the ballpark call.

Did it work?

PageMotor keeps this site free.