How I Built a Real App with AI in 10 Days — And What Nobody Tells You
I Built MAKAN Translate with AI. The Hardest Part Wasn't Writing Code. Subtitle: Four development approaches, dozens of model experiments, one abandoned game translator, and the engineering lessons behind a free multilingual application. Makan · Founder, MAKAN LAB · October 2026 Type: Independent engineering case study / reflective technical report Project: https://makanlab.tech/translate/ ·…
MAKAN Translate was born from a need to help a friend study Linux using translated Udemy subtitles. The software began as a simple interface connecting to a translation API, but it quickly became clear that this approach had limitations. The first major experiment took place in Google AI Studio, where an impressive-looking application was generated with just a few clicks.
However, the underlying product proved to be unreliable, and the same issue appeared in subsequent experiments with ChatGPT/Codex and Claude Code. The author highlighted the importance of task decomposition and validation in the development process, emphasizing that increasingly elaborate prompts or the selection of a single coding assistant did not solve the core problems.
The project ultimately involved a division of research, review, coding, and manual verification across different tools, revealing the engineering lessons behind a free multilingual application.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
