Ra9mana: Turning Excel Sheets and Paper into a Business App with AI
An AI-assisted digitization platform that reads a company's existing Excel files and documents and generates a bilingual operational web app around how they already work.
Laravel 12ReactPostgreSQLAI
Context
Many Moroccan SMEs run on scattered spreadsheets, paper delivery notes, and WhatsApp messages. Off-the-shelf ERPs force rigid accounting workflows onto operational teams, while custom agency builds take months — so teams abandon new tools after two weeks and fall back to Excel.
Scope I disclose publicly
- An ingestion engine that analyzes existing Excel workbooks, CSV exports, scanned invoices, and process descriptions to infer entities, relationships, and validation rules.
- A Laravel 12 and React runtime that turns the inferred schema into an Arabic/French/English workspace with role-based views and approvals.
- A PostgreSQL data layer with lossless Excel import/export round-tripping so operators never lose access to their spreadsheets.
- Operational modules for daily cash boxes, supplier balances, inventory movements, and audit logs.
Decisions and trade-offs
- Schema-first AI generation: LLMs propose typed schemas and workflow rules rather than raw unmaintainable code, keeping every workspace deterministic.
- Treating Excel compatibility as a first-class feature rather than an afterthought, allowing gradual adoption without data lock-in.
- The trade-off: constraining generation to structured operational primitives limits arbitrary UI layout freedom in exchange for predictable data integrity.
Ra9mana is an active internal product in a private repository; live architectural walkthroughs are available during technical calls.
Disclosure: I do not publish pilot company names or financial ROI figures. This page describes the engineering architecture and built scope only.