SSS — Sarena Shooting Star

Small shop and restaurant owners in Cameroon and similar markets run their business on WhatsApp, in notebooks, and with borrowed designers. SSS replaces the patchwork with one platform that respects how they actually work.

The product is built around four engines. SASTO handles the business: products with SKUs, barcodes and suppliers; stock tracked through a full movement ledger so every level is traceable; sales that create invoices and deduct stock only on payment or fulfillment, never twice; clients, orders, notes and warnings. Teams have clear roles — owner, manager, staff — with permissions enforced everywhere.

AIcSA manages brand identity through a STYLE.md system that learns from uploaded flyers or a logo, then generates reusable design templates. The Illustration Maker fills these templates with new prices or text without burning AI tokens again. Owners can edit, remix, remove backgrounds, crop, adjust colours, and export to PDF or PNG. Web lookup is optional; any image used is saved locally, never hotlinked.

SATBO and SAWABO are Telegram and WhatsApp bots. Customers message naturally; the assistant answers from real catalogue data, sends photos, and takes orders. Business owners run operations by chat — "Sold 3 to Awa, paid" records the sale, updates stock, creates the invoice. The AI proposes actions but never owns data; the business engine validates and writes everything.

A public shop page at `/b/<slug>` gives each business a website with no code: hero, products, services, promotions, gallery, hours, contact form. QR codes and shareable links feed leads back into the app.

The assistant works on the web and Telegram, with WhatsApp built but held until a product number is available. It respects roles, confirms risky actions, and burns no credits on unknown messengers.

Billing is in FCFA via Orange Money and MTN MoMo. Plans split AI credits between chat (Kimi K2.6) and design (Claude Sonnet 5.5); when a pool runs out, that feature pauses until the next cycle. Admins get spend tracking in USD and FCFA, business management, announcements, and group controls.

I architected and built the full stack: Next.js frontend on Vercel, API on a dedicated server, bots containerised in Docker, AI orchestration with token budgeting, and the design pipeline with template caching.