I run a small SaaS-based trading ERP that serves 10 client companies managing roughly 600 customer accounts. All payment details—transactional history, scheduled instalments, and any outstanding invoices—are already stored in our database. I need an AI-driven module that analyses this data, flags late-paying customers in real time, and segments every account by payment reliability so my users can follow up sooner and smarter. Scope of work • Build a RESTful API (JSON) that connects to our existing ERP database and accounting software. • Ingest the three data streams above, run the analysis, and return: – A late-payment flag (yes/no, days overdue). – A reliability score or tier (e.g., A–D) based on historical behaviour. - Different Customers have different payment patterns. • Expose endpoints for: on-demand customer lookup, bulk nightly refresh, and summary dashboards. • Provide a straightforward authentication method (token-based) that we can secure behind our gateway. • Supply a short setup guide plus sample Python/JavaScript calls so our in-house team can plug it straight in. I’m looking for a clear, well-commented codebase, a concise model training notebook (if ML is used), and basic success metrics to show the approach is working on our sample data. COST CONCERNS ARE THERE. KINDLY LET ME KNOW PROPER PLAN AND WILL IT INCUR RECURRING COSTS?