Technology

AI and Honeylogo

AI and Honeylogo

In the 1980s, as a teenager, I developed a Logo implementation for the BBC Micro, which was widely used in UK schools.

Recently, I was asked to resuscitate that code. On review, it was an impenetrable mess of GOTO statements, with barely any newlines. Trying to restore it was pointless.

I turned to my programming assistant AI, Windsurf, for help. My first attempt used Golang, but completing it required too much effort. I tried again—this time with TypeScript and React. I asked the AI to convert the Golang code to TypeScript. The conversion was rapid: within three hours I had gone from zero TypeScript to a deployed app with a registered domain.

Because Logo is a known language, the AI could leverage prior understanding. I gave it a list of commands and a syntax document from another Logo implementation. It responded by implementing 23 commands, using its own interpretation of the language structure.

People often debate whether AI is transformative or useless. For me, it was mindblowing—seeing a decades-old, broken mess transformed into a working app in hours, thanks to an AI that understood the domain without handholding.

In total, I’ve spent about three or four days on the project. Without AI, it would’ve taken a month—just the typing alone would have consumed days. This project wouldn’t have happened without it.

HoneyLogo will continue to improve. Print materials are being developed to support users. Watch HoneyLogo for updates.

If you have feature requests, please send them via the Contact page on this website.

Join My Newsletter

Receive updates when I have something meaningful to share.