Chatac.ai

Chatac.ai is a full-stack educational platform built on Rails 8 that automates course creation through AI. Instructors upload source materials, either PDFs or videos, which are processed via OCR and converted into semantic chunks. A RAG pipeline then generates a hierarchical mind map, and an AI module produces micro-lessons organized around 7 pedagogical learning vectors.

One of the most deliberate decisions in building Chatac was around the frontend stack. We started with React. It gave us flexibility and a familiar component model, but over time it created a clear split between the frontend layer and a Rails backend that was already doing all the hard work. Context switching between two different paradigms slowed us down and introduced duplication we did not need.

We moved the entire UI to Hotwire, keeping everything inside the Rails ecosystem. The result was a simpler codebase and a team that could ship faster. Real-time interactivity was handled through ActionCable and Turbo Streams, which fit naturally into the same server-rendered flow. The whole stack became one coherent system instead of two applications exchanging data.

Migrating the UI to Hotwire required rethinking how the interface consumed server state. It was uncomfortable at first, but the payoff was clear.

On the AI side, producing consistent structured outputs from the model required careful constraints, including capping quiz answer options at 20 characters for WhatsApp compatibility, while keeping the experience responsive through Turbo Stream updates during long-running background jobs.

Full ownership of an AI-first product taught me to question architectural choices at every stage. The shift from React to Hotwire reinforced something important: the best technology is often the one that fits the ecosystem you are already in.

Rails with Hotwire is capable of powering real-time, AI-driven products without a separate frontend framework. That realization changed how I approach new projects.