AI Lab
A coding agent at work
This is what AI-native software development looks like: an autonomous agent receives a task, inspects the code, writes tests and delivers the solution. This demonstration shows a typical run – step by step.
Agent working …
Agent
The checkout test is failing – find and fix the root cause.
Workspace
…
What you are seeing
The agent works in a loop of reasoning, tool use and observation: it reads code, forms hypotheses, verifies them with tests and corrects itself – until the task is demonstrably solved.
This is exactly how I use LLM agents in real projects: with clear tasks, verifiable results and an experienced engineer owning architecture and quality. Through workshops and training I bring this way of working into your team as well. The demo runs entirely in your browser – no external AI calls.