As humans, our advantage over AI is our life experiences which compound and accumulate over time.
For many years, I have been an avid journaler. Stream-of-consciousness writing has become one of my primary ways of thinking strategically, generating ideas, and often working through solutions that eventually become projects.
Moving from physical journals to digital systems such as Obsidian is something I have been practicing for the last 10+ years, although I still very much enjoy a physical notebook.
With that said, the advent of AI has supercharged the value of journaling.
It turns out that this same practice is also required for organizations to achieve the promised gains of AI and agentic systems. I wrote about this earlier in Documentation Is Becoming Infrastructure.
From journaling to infrastructure
It has become a personal ambition to continuously enhance my own capabilities. Every meaningful life experience and every project generates information and useful knowledge.
When we intentionally capture that knowledge inside our own personal knowledge management system, and combine it with a personal agentic operating system, new opportunities and capabilities emerge.
This is the future. Actually, it is already here.
The Knowledge Engine
I have been building my agentic system to harvest my personal experiences and knowledge. Some of it comes from technical decisions, research, strategy, writing, leadership, and the work of solving problems with other people.
Some of it gets captured in documents, code, notes, and conversations.
Just as with journaling, collecting information is not the hard part. The challenge is keeping that information organized, connected, current, and useful long after the experience, the project, or the timeframe has passed.
Knowledge gets scattered across notes, repositories, documents, chats, and drafts. Some knowledge has been reviewed enough to reuse, some is still a working idea, and some only made sense inside the situation where it was created. But this knowledge is available for agent evaluation and selective curation into new artifacts, directed and governed by me.
Proprietary intelligence
Bain & Company recently wrote about turning artificial intelligence into proprietary intelligence. The real advantage is combining AI with accumulated knowledge, context, workflows, and continuous learning. This proprietary intelligence is the moat for organizations, and will be what sets humans apart from every other human who has access to the same AI.
Infrastructure has to be maintained, reviewed, and owned. Its purpose and history need to be visible and accessible.
I think knowledge deserves the same treatment, especially when I want to use AI agents to help me work across it.
This is the premise behind my personal Knowledge Engine.
I built a private system that I continue to evolve and trust with my own work. The goal is to capture what I am learning, connect related ideas, preserve context, and improve that knowledge and the system over time.
This is also how I stay deeply engaged at the forefront of AI capabilities: by building, evaluating, implementing, and learning the different aspects of agentic engineering.
The public pattern language
I built Agentic Operating Patterns as the public view of that work, with the private system, agents, configuration, and implementation details extracted.
It is the public pattern language coming out of my private system. It explains how knowledge is captured, curated, retrieved, reviewed, and governed under my direction, without exposing the private system underneath it.
If people are going to collaborate with AI agents on serious work, the knowledge underneath that collaboration has to be maintained like infrastructure that supports a production system.
In this case, an agentic operating system.