Concrete examples to multiply agents and open new use cases — all built by talking to Claude. Copy the prompts, tweak the words, and adapt them to your client.
Reminds, confirms, and warms up the lead before the call. Cuts no-shows.
"Create a pre-call agent based on my existing setting agent: it reminds and confirms the call, cuts no-shows, and warms up the lead before the call. Reuse its voice."
Revives the dormant base on WhatsApp with a no-pitch opener and books.
"Build me a WhatsApp re-engagement agent for my cold base: a no-price opener, identifies as AI, and books a diagnosis call."
Answers questions, onboarding, and post-sale — frees up the human team.
"Create a support agent that answers FAQs and client onboarding, and escalates to a human when it doesn't know."
Replicate the one that works on Facebook or TikTok, reusing everything.
"Replicate my Instagram agent on Facebook and TikTok, reusing voice, examples, and keywords, and leave it measured."
Pull the CSV from your CRM (GHL, etc.). Ask Claude to filter: exclude buyers, DNC ("don't message me again"), and anyone who canceled less than 14 days ago. A small clean list beats coverage.
Headers: First Name · Last Name · Email · WhatsApp Number · Agent Context. The number goes with + and digits. Max 2000 rows per import (split it if larger).
Tag each cohort (e.g. top5k_warm, country, "booked before yes/no"). That lets you measure per segment and branch the follow-up. Ask Claude to build the balanced segments.
The first outbound message must be an approved template (HSM). To test A/B/C/D, approve one template per variant — otherwise they all go out with the same one.
The template goes out; when the lead replies, the 24 h window opens and the agent chats freely, qualifies, and sends the booking link. Outside 24 h, approved templates only.
Real case: re-engaging Esculpe's base (~4,900 warm leads, 4 segments, agent "Angie WA - Recontacto"). Lesson: subtract buyers and the cohort that already bought before firing, or you cold-message current clients.
Let Claude read your export and build cohorts ready to fire.
"Take this CRM CSV, subtract buyers and DNC, and build me 4 balanced segments by country and by whether they booked before, in Ninjō Studio format."
Cross your base with purchases and bookings so you don't cold-message the wrong people.
"Cross my warm list with buyers.csv and the calendar, and flag who already bought and who already booked before."