Four-person landscaping crew drafts same-day estimates from site-visit voice memos with Claude
Four-person landscaping company (author's own) · Ontario, Canada ·
What they did
The office manager records the site visit on her phone, pastes the transcript into Claude with specific pricing logic, and reviews the draft estimate, correcting regional pricing and warranty wording before sending it the same day. Generic prompts produced "beige nonsense". It only became useful once real pricing rules were in the prompt and Claude was told never to invent line items. Rolling out four workflows in one week failed, so they went back to one at a time.
Reported results
- After 90 days: admin time dropped from about 16 to about 6.5 hours a week, estimate turnaround went from 3–5 days to same day, and close rate on jobs under $10K rose from about 40% to about 58%
Tools
Claude
Cost
About $20/month in API costs (as reported by the owner)
Source
Contractor Tool Shop: https://contractortoolshop.com/blog/90-days-of-ai-on-a-landscaping-crew
Builder write-up. Written by the owner. The source publishes the exact estimate prompt it used. Source checked 2026-10-20. Results are as reported by the source, not independently audited.
Try it yourself: starter prompt
Our suggested prompt for an AI assistant (ChatGPT, Claude, Grok or your phone-agent tool). It is not from the source.
Here are my notes and photos from a site visit for a landscaping company: [paste notes]. Using my rate sheet [paste rates], write a clear, friendly quote with line items, what's included, what's not, and a validity date. Flag anything not on the rate sheet for me to price.