I’d been growing frustrated with AI assistants for months before I realized what the problem was. I’d been using various AI coding tools (GitHub Copilot with Claude Sonnet and Opus models, occasionally trying Gemini and GPT variants, plus Microsoft 365 Copilot for one-off tasks like image generation and brainstorming project names), and while they were helpful, they all followed the same limiting pattern: type a prompt, get a response, move on. Most LLM chatbots can remember things from previous conversations now, and even perform actions on your behalf, like writing and running code or searching the web, but they’re all fundamentally the same in one critical way: they just sit there and wait for you to talk to them.

If you’ve spent any significant amount of time working with AI agents, you’ve probably run into a frustrating limitation: they can remember individual facts between sessions, but they have no understanding of how those facts relate to each other. It’s a lot like Leonard Shelby in Memento, who literally marks down his memories on paper, polaroid photos, and tattoos. He can read any individual memory and know what it says, but he can’t piece together how they all connect. That’s basically what your AI agent is doing with its markdown memory files: storing a flat collection of facts with no relational structure between them. It works for simple use cases, but it falls apart once your project’s knowledge starts looking more like a web than a list. So, I’ve been building something I think is fundamentally better. I call it Axons for Agents: a graph-based memory system modeled after how human brains actually store and retrieve information, with brain-like plasticity, memory compartmentalization, and an MCP server that lets AI agents use it as a native tool.

This is the sixth installment in a series documenting the challenges, progress, setbacks, and victories of The Data Dojo: A Power BI Community of Practice. In this post, we’ll talk about a brand new style of Data Dojo workshop, which we call “Office Hours & Coffee Lounge,” why we decided to add this new format to our repertoire, and how it’s been going so far.

This is the fifth installment in a series documenting the challenges, progress, setbacks, and victories of The Data Dojo: A Power BI Community of Practice. In this post, we’ll take a closer look at the latest Data Dojo template, tinker with some of its capabilities, and talk about how to strike the right balance when designing a Power BI template that’s both simple enough for a beginner and versatile enough for a seasoned pro.

This is the fourth installment in a series documenting the challenges, progress, setbacks, and victories of The Data Dojo: A Power BI Community of Practice. In this post, I’ll tell you about our 4th and 5th workshops (our most important and impactful workshops yet), and I’ll explain why I believe that every Power BI Community of Practice should host workshops like these as often as possible.