Cheezy World

Building an AI-Optimized Kanban System: A Human-AI Collaboration Experiment

I’m embarking on an experiment to build a Kanban task management system specifically designed for AI-human collaboration. This isn’t just another project management tool—it’s a system where AI agents and humans work together as peers on software development tasks.

The Vision

Most task management tools treat tasks as simple items with a title and description. But when you’re working with AI agents (like Claude, GPT-4, or custom agents), they need much more context to be effective. They need to understand:

Bringing AI into your organization

I have now worked with multiple teams that have introduced an AI driven development into their development workflow and I have seen a pattern emerge. When AI, and more specifically Vibe Coding, is introduced into a traditional software development organization that are following an Agile process there are three tensions that quickly arise. All three of these tensions are due to the rapid pace of development.

I have made my living as an Agile coach since 2003. I have worked as a coach for clients around the world, taught many courses in various topics, and spoken at too many conferences to count. I have dedicated more than a third of my life to these concepts and truly believe they have been the best way to deliver software.

Agile has gone through many changes over my career. A few of the major ones that I feel have had the most impact are the introduction of Lean Startup, practices like BDD and Acceptance Test Driven Development, and my favourite which is Continuous Delivery. Most of these changes have come from within the Agile community and have pushed us forward in our understanding of how to deliver more value quicker. I have stayed current and in some cases have been on the forefront for many of these changes.