Quote Follow-Up Sequencer (Universal)

Any business that sends quotes and watches a significant percentage go cold. Industry-wide, 60-75% of quotes never convert, most of those because the customer "needed to think about it" and never got followed up with. Particularly powerful for: roofing, HVAC, plumbing, kitchen/bath remodeling, landscape, painting, solar, pool construction, auto body.

Also a perfect CROSS-SELL to any client already using one of the visualizer tools in this library (Roofing Satellite Quote, Pool Backyard Visualizer, Kitchen Re-Render, etc.), those tools generate quotes; this tool stops them from dying.

The business owner adds a quote to the system (manually entered, imported from QuickBooks/ServiceTitan/Jobber, or auto-imported from any of the visualizer tools in this library when bundled). The system triggers a multi-touch follow-up sequence over 14-30 days depending on the vertical: a Day 2 check-in, a Day 5 value-add message, a Day 10 social proof touch, a Day 14 limited-time urgency, a Day 21 final close, and a Day 30 graceful exit. Each message is generated by an LLM personalized to the customer name, quote amount, and service type.

The sequence INSTANTLY HALTS the moment the customer replies anything, so the owner never accidentally messages someone who's already booking. Owner gets SMS notification of every reply with classification (booking/question/declined/complaint).

This is the second-most-complex tool in the library (after AI Receptionist). The trickiest piece is the sequence engine, it must atomically halt sequences on customer reply with NO race conditions. A message firing AT THE SAME TIME the customer replies is catastrophic.

Expect Base44 to build in 15-20 minutes. Likely follow-up prompts:

If sequences keep sending after replies: "The sequence_halted check isn't atomic. Race conditions are letting messages send to customers who've already replied. Before sending ANY scheduled message, re-fetch the Quote record from the database and check sequence_halted = false.

If true, mark the MessageScheduled as 'skipped_replied' and exit. This check must happen INSIDE the send transaction, not before."

If the LLM is generating awkward touch numbering ("Wanted to follow up for the 4th time..."): "The LLM is referencing the touch number explicitly in the message body. Update the prompt: 'Do NOT mention the touch number, the sequence, or that this is an automated follow-up. Each message should feel like a fresh, organic check-in.'"

If sequences aren't auto-importing from a sibling Base44 visualizer tool: "The importFromSharedWorkspace function isn't picking up new leads from the linked visualizer tool. Check: (1) the linked_visualizer_tool_workspace_id is set correctly in Settings, (2) the function has read access to the visualizer tool's Lead entity, (3) the scheduled function is actually running every 15 minutes, check the Functions panel logs."

If timing is wrong for international customers: "The send_hours check is firing in server time, not customer time. Geocode the customer's address at quote creation time, store their timezone, and check business hours in their local timezone before sending."

The biggest gotcha: explaining to the business owner that auto-halt is non-negotiable. Some owners want to "manually approve each touch", but that defeats the purpose. The system is set-and-forget once approved.

They review replies in the Inbox tab, not every outbound message.