Reimagining Backend Development with AI-Powered Execution15 Feb 2026
My first experimented with an AI agent that could generate a backend API using the NAIV framework, but the process was slow and costly due to repeated LLM interactions. In the second attempt, I created the Flazy API engine, which dynamically generates, transpiles, and executes API function code on demand. This approach allowed the system to handle requests immediately, even before the API functions were fully implemented, demonstrating surprising reliability. In the third attempt, the focus shifted to validating AI-generated code rather than writing it, marking a move toward a fully autonomous, reinforcement-based API engine powering the website itself.
From a Hobby Project to AI-Powered Development14 Feb 2026
In 2021, I introduced my hobby project, VISP (Visual Programming), a simple web interface for defining functions through graphical relations, but it never gained traction. Frustrated by repetitive coding tasks in my development career, I experimented with simplifying my approach, leading to the creation of tools like Pagonila. In 2025, I developed NAIV, a code generation tool that finally realized my vision of simplifying backend and frontend development. By 2026, I began exploring AI-assisted coding, aiming to create a structured, AI-guided development process for faster, more reliable software creation.