The AI Reset
A free video course. Fourteen modules take you from how LLMs actually work to agents, MCP, and what it all means for your company. No signup. Nobody sells you anything.
Start at module one. Watch in order. By the end, you will understand AI better than most of the people advising you on it.
- Introduction
- How LLMs work
- Customizing LLMs
- AI agents
- Web apps
- Model Context Protocol
- Understanding AI
- AI infrastructure
- AI impacts
- AI risks
- Prompt engineering
- Working with large files
- Drop everything
- Where to go from here
Who reads page four in your office?
A fund buried a $12,000 fee increase on page four of an amended agreement. No human caught it. An agent that reads every incoming document did.
I wrote The Agentic Family Office because I run agents in my own office, and I know where they break. The book stays skeptical the whole way through. Forty-two chapters take you from what an agent actually is, through models, memory, security, and giving an agent its own card, to a road map you can run in phases. I also tell you where plain software still beats an agent, because it often does.
The book costs nothing. Start with chapter 30, where I show what goes wrong, then read the road map in Part VIII and decide how fast your office should move.
- What agents are and what they can do
- How agents work
- The brain: models, memory, accuracy
- Harnesses
- Infrastructure: local, cloud, hybrid
- Operations: privacy, money, governance
- Case studies
- The big picture
You think you have AI covered.
You have AI people, maybe a whole department. You run agents in production. You hold a strategy meeting every month. But everyone in that room reads the same feeds, hears the same vendors, and shares the same assumptions. The room agrees with itself.
I bring the view from outside. I spend my days building with agents, testing the tools, reading the research, and questioning the claims. Once a month I tell you what actually happened in AI, what is real and what is hype, and what is about to hit your industry. Then we turn to your situation.
The traps are expensive. You back a platform your team just approved, and the model companies give it away six months later. You miss the shift that reprices your market. I see more than you do, and two heads beat one.
Learn agents by building agents.
I open with a short talk. I show you what agents are, where they beat ordinary software, and where they quietly fail. Then your team takes the keyboard.
Everyone picks a track and builds something real on your own workflows and your own content, while I work alongside them. Your people leave with a pilot they built, not slides they watched.
Your team walks away with a working pilot, a framework for deciding what to automate, and the judgment to know when an agent is the wrong tool. The capability stays after I leave.
Everything in the learning center, newest first.