A Veteran Programmer and an AI Built a New Language in a Day

Last December, in a quiet Brussels workshop, software engineer Bernard Lambeau began a unique experiment. Instead of writing code himself, he presented an idea to the AI assistant Claude. Within 24 hours, they had created Elo—a fully functional programming language with compilers, a type checker, and complete documentation.

Lambeau, a PhD-holding founder with three decades of experience, directed the project. Claude wrote every line. The result, as detailed in a joint blog post and later reported by The Register, is a portable system that compiles to JavaScript, Ruby, and SQL.

Elo is designed for safety and simplicity, specifically for use within no-code platforms. It prevents common security exploits by design, eliminating mutable state and side effects. Its straightforward syntax allows non-programmers to build logic, like filtering customer lists or checking trial expirations, without traditional coding complexity.

Lambeau’s methodology was systematic. He began by supervising Claude closely, then granted the AI autonomy within isolated development environments. Claude wrote its own tests, executed them, and corrected errors. The entire process is documented in over 100 task files on the project’s GitHub repository.

“Claude Code knows almost every tech stack and works about ten times faster than I can,” Lambeau noted. He estimates the project would have taken weeks alone, or months with a hired developer. Instead, it cost the monthly subscription fee for Claude’s advanced tier.

Elo is already in use at Lambeau’s company, Klaro Cards, allowing non-technical users to safely create custom business rules. The project joins other recent experiments where experienced developers use AI as an implementation partner, dramatically accelerating development cycles.

Lambeau emphasizes that this approach requires deep expertise to guide the AI effectively. For those who possess it, however, the collaboration points toward a new model of software creation, where human architects and AI implementers build together.

Source: Webpronews