SoftifyAI – AI Corporate Softener.
A Next.js application that rewrites angry or informal messages into polished corporate communications using multiple LLM providers.
The objective
People need help turning blunt or emotional messages into professional corporate communication quickly.
Architecture & implementation
Built a Next.js SPA with multi-provider LLM fallback, SQLite reply caching, Hugging Face embeddings warm-up, and PM2 production deployment.
- User
- Next.js App
- SQLite Cache
- LLM Providers
- Hugging Face Embeddings
Key components of the implementation.
My contributions & decisions
Multiple model providers
Implemented LLM fallback across OpenCode, OpenAI, Groq, Kimi, and Gemini for rewriting messages into professional communications.
Caching and model preparation
Added SQLite reply caching and Hugging Face embeddings warm-up.
Production deployment
Built the Next.js application with TypeScript and React, and deployed it using Node.js and PM2.
The outcome
Deployed live at softifyai.mrox.in.