What the agent did
Claude, connected to Zapier through Zapier MCP, writes and publishes Zapier workflows from Ross's descriptions. The running workflows are scheduled Zaps that pull ServiceTitan data, flag bookings pending 20 minutes, send a 6:30 a.m. weekday brief and send matched calls and job revenue to Google Ads.
Result, as claimed
Zapier reports 36 live Next-Gen Zaps running about 1,400 times a week across about 30 connected apps. Revenue went from about $300,000 a year to a projected $3.8 million in 2026, but neither source attributes that growth to AI.
This is the result as claimed in the public source linked below. It has not been checked and is not a typical outcome.
Proof
Try it yourself
Use Claude to find leads nobody has called back, from a simple export.
Prompt
I run [business type]. I will paste an export of enquiries from [CRM or booking system] with columns: name, date and time received, status, assigned person. Today is [date and time]. List every enquiry still marked [pending status] for more than [20 minutes or your rule], oldest first, with how long it has waited. Then list any enquiry pending more than 24 hours separately as Urgent. Do not contact anyone. Use first names only in your output. If a column is missing or unclear, say so instead of guessing. Return a draft only. Never send, spend, change settings or write to a live system.
Prompt: starter by Solopreneur Code
Tools
- Claude Source-named tool. Official product URL not independently verified during intake, omitted.
- Zapier MCP Source-named tool. Official product URL not independently verified during intake, omitted.
- ServiceTitan, Google Ads Source-named tool. Official product URL not independently verified during intake, omitted.
Steps
- Pick your response-time rule, for example 20 minutes.
- Export today's enquiries with received time and status.
- Run the prompt and call back the oldest leads first.
- Run it daily for a week and note any leads it missed.
- Only then describe the check to Claude as a Zapier workflow and test it on old data before turning it on.
Watch out: Exports often store times in a different time zone. Check one or two lead ages by hand before trusting the list.
Starter version by Solopreneur Code