(CATEGORY)

AI engineering

This category is about how an AI that writes and reviews programs really works in practice. In the article on controlling an AI in production I explain three mechanisms that keep mistakes away: rules that say in advance when a job is finished, a check of every finding against the program as it is today, and an automatic return to the previous good version when a change fails. In the article on the Content Studio of Praion AI I show how text, images and video for a business’s posts come from different models and always pass through the approval of a person. The articles are written in plain words, for readers without a technical background, and each of them starts with the answer. The category is for anyone who wants to understand what an AI project needs before trusting it inside their business. This page lists the 2 articles in the «AI engineering» category, written from real work and sorted from the most recent. Each article starts with the answer and continues with the facts behind it.

Articles: AI engineering

Illustration: an AI assistant passes three checks before going live, with a way back

AI engineeringSeptember 2026 · 8 min read

How to keep an AI agent that writes production code under control

The model is not the guarantee. The guarantee is three things: completion criteria that leave no room for interpretation, verification of every finding before it is accepted, and automatic rollback when a deployment fails. What I learned building Planner.