Home

Sara3com

Next.js 16TypeScriptPrismaPostgreSQLNextAuthOpenAIGeminiDocker

Overview

Sara3com is a full-stack telecom company website built for an Angolan telecommunications provider. The platform serves as the company's digital presence with the tagline 'Sua Conexão, Nossa Inovação' — featuring service management, customer authentication, a newsletter system, and a complete admin dashboard.

The standout feature is Sara AI, an intelligent chat assistant powered by 7 different AI providers (OpenAI, Google Gemini, Cohere, Mistral, DeepSeek, XAI, and Google Generative AI) with a dynamic model registry that allows seamless switching between providers.

Sara3com preview
Sara3com detail

Key Features

  • Multi-provider AI chat assistant (Sara AI) with 7 integrated LLM providers and tool-calling capabilities
  • Admin dashboard with full CRUD for services, users, and content management
  • NextAuth-based authentication with Google and GitHub OAuth providers
  • Newsletter system with Resend email integration and QStash background job processing
  • Service catalog with dynamic detail pages
  • Prisma ORM with PostgreSQL (Neon serverless) for data persistence
  • Docker containerization for deployment flexibility
  • SEO-optimized with next-sitemap and PWA manifest

Technical Approach

Built on Next.js App Router with a clean separation between public pages, authenticated routes, and admin-only sections. The AI chat system uses a modular architecture where each provider is registered in a central config, allowing runtime model selection.

Authentication is split into read-only and adapter configurations for optimal Next.js compatibility. The admin dashboard uses tab-based navigation with dedicated components for each management section.

The project follows a services-based architecture with dedicated API service layers for data fetching and auth, React context providers for theme and session state, and a hooks layer for shared UI logic.

Outcomes

  • Deployed to production, serving as the company's live website
  • 96.6% TypeScript codebase ensuring type safety across the entire application
  • Successfully integrated 7 AI providers into a unified chat experience
  • Built and shipped in 26 commits with a clean, production-ready codebase