Service Reminder Engine (Universal)

Any local business with a recurring-service model and an existing customer list they're not actively re-marketing to. The ideal pitch target is a business that's been operating for 3+ years and has hundreds or thousands of past customers sitting in QuickBooks, Service Titan, Jobber, a paper folder, or just the owner's memory. Every one of those customers is worth $500-5,000 in future revenue and the business is leaving it all on the table because they have no system.

This tool IS the system.

The business owner uploads their customer list (CSV from QuickBooks, Service Titan, Jobber, Excel, or even a hand-typed list). The AI parses the list, classifies each customer by their last service type and date, predicts their next-due date based on vertical-specific re-service cadences, and generates a personalized SMS for each customer at the right moment. Messages reference the customer by name, their specific service history, and book them with a one-tap reply.

The business owner reviews and approves the messages in batches before they send, then watches the booking dashboard fill up. This tool unlocks revenue from data the business already owns but isn't using.

This tool is mechanically less complex than AI Receptionist but has unique challenges around data quality and compliance.

Expect Base44 to build in 12-18 minutes. Likely follow-up prompts you'll need:

If the CSV parser is choking on different formats: "The CSV parser is failing on common formats. Add robust handling: detect column headers automatically (look for variations like 'name', 'customer', 'client', 'first_name + last_name'), handle date formats including ISO, US, EU, and natural language ('March 15, 2024'), handle phone formats including parentheses, dashes, spaces, with or without country code. Use a forgiving parsing library like Papa Parse for CSV and chrono-node for dates."

If the LLM is generating messages that sound robotic: "The generated messages sound too generic and corporate. Adjust the LLM prompt to be more directive: 'Write like a small business owner texting from his phone. Use sentence fragments.

Don't use formal language. Read like a friendly text, not an email.' Also include 3-5 example outputs in the prompt to show the style we want."

If the rate limiting isn't kicking in properly: "Messages are sending too fast and triggering Twilio carrier filtering. Strictly enforce 20 messages per minute via a queue with timestamps. Each function call to sendBatch should pull only 20 messages at most, wait the rest of the minute, then exit.

Use Base44 scheduled functions to call sendBatch every minute."

If opt-outs aren't being handled: "Opt-out replies aren't being detected. Add aggressive STOP keyword matching: STOP, UNSUBSCRIBE, REMOVE, QUIT, CANCEL, OFF, NO, END. Match case-insensitive.

Add the phone to OptOutList immediately. Send the confirmation reply. NEVER send another marketing message to that number ever again, regardless of which campaign it's part of."

The biggest gotcha: A2P 10DLC registration. Without it, the client's Twilio account will get throttled aggressively by carriers after 50-200 SMS sends. The Phase 2 playbook walks through this.

Students MUST complete A2P 10DLC for the client BEFORE launching the first real campaign.