Serve Aí: plataforma multi-tenant para autoatendimento e gestão no setor gastronômico

Abstract

Este trabalho documenta o desenvolvimento da plataforma web Serve Aí, solução que integra autoatendimento digital e gestão operacional para estabelecimentos gastronômicos. O sistema utiliza arquitetura multi-tenant, permitindo que diferentes estabelecimentos compartilhem a mesma aplicação mantendo isolamento completo de dados, seguindo a taxonomia de Chong e Carraro (2006). A base técnica compreende dois módulos desenvolvidos com Server-Side Rendering (SSR), React Server Components e TypeScript: interface de autoatendimento para clientes finais e painel administrativo para gestores. O desenvolvimento aplicou padrões MVC e Repository Pattern, com segurança orientada pelas diretrizes OWASP. O módulo cliente utiliza Teoria da Carga Cognitiva (Sweller, 1988) e Progressive Disclosure para otimizar conversão. O módulo administrativo oferece gestão de pedidos em tempo real, histórico paginado e integração com Power BI. A validação técnica demonstrou que a combinação de renderização no servidor com isolamento de tenants atende requisitos divergentes de simplicidade (consumidores) e densidade informacional (gestores), viabilizando sistemas com alta performance e escalabilidade. A arquitetura permite expansão futura com assistentes virtuais e sistemas de recomendação baseados em inteligência artificial.

This work presents the development of a multi-tenant web platform named Serve Aí, designed to provide an integrated solution for self-service and operational management in food service establishments. The project is grounded in the application of modern software architectures, combining Server-Side Rendering via React Server Components with TypeScript to build two functionally segregated modules: a self service module for end consumers and an administrative panel for operational management. The implemented multi-tenant architecture allows multiple establishments to share the same application instance while maintaining rigorous logical data isolation, following Chong and Carraro's taxonomy (2006). The methodology was based on component-oriented development principles, applying established design patterns such as MVC and Repository Pattern, alongside OWASP security guidelines to protect against access control vulnerabilities. The client module implements conversion optimization strategies grounded in Sweller's Cognitive Load Theory (1988) and Progressive Disclosure principles, prioritizing reduced Time to Interactive through server-side rendering. The administrative module incorporates real-time order management functionalities, server-side paginated operation history, digital menu management, and integration with analytical dashboards via Power BI. Technical validation demonstrated the viability of the proposed dual architecture, simultaneously meeting divergent requirements of simplicity for consumers and informational density for managers. Results confirm that the application of Server-Side Rendering with multi-tenant isolation constitutes an effective solution for systems demanding high performance, scalability, and clear separation of responsibilities, establishing an extensible architectural foundation for future functionalities such as virtual assistants and artificial intelligence-based recommendation systems.