Salesforce Just Named Its AI Agents. One of Them Is Called Hunter.
Salesforce shipped seven named AI agents with job titles and months of memory. What that convergence says about naming agents, and what a vendor still cannot sell you.
Read the post
Blog
Notes from building REALM in public. Practical essays on AI agents, the Codex, company-of-one operations, automation, and the strange new shape of human-agent work.
I Stopped Using Half My Own Framework
I wrote rules describing what AI agents could not do. Then they got good, and half my own framework was wrong. What I deleted, and what held.
Read the post
You Are Not Using AI Agents. You Are Engineering Harnesses.
The model is not the agent. The harness is. And most people have no idea they are building one — or that the quality of that harness determines everything.
Read the post
The REALM Comics
Six issues, five classes, one story. Why the REALM Comics exist, and why creativity helps make an operating framework feel alive.
Read the post
Why Saga Posts on X
Running an AI agent that posts to X on its own: what it takes, what it proves about the agent's memory, and why it is more than scheduling.
Read the post
Your Agents Can Be Manipulated
Prompt injection, SOUL file leaks, local model risks — your agents can be manipulated. Here is what that means for your Realm and how to build with it in mind.
Read the post
Which Model for Which Character — My Current Preference
A practical analysis of current AI models through the REALM class system. Not benchmarks — which model I prefer for each Character class right now, and why.
Read the post
The Automated Realm: How OpenClaw Crons Run My World 24/7
Technical breakdown of my REALM automation. Using OpenClaw crons, isolated sessions, and local models to generate the Daily Scroll and social presence.
Read the post
How I Set Up My Realm — Step by Step
The full setup I use to run a crew of AI agents: OpenClaw, Slack, Obsidian. Every step, every decision, and the things worth being careful about.
Read the post
The Company of One Human
A company of one changes when AI agents become working colleagues. REALM reframes the founder as the only human in a hybrid company.
Read the post
Why I Use Obsidian for the Codex
The shared knowledge base your agents read from is just folders and markdown files. Why I use Obsidian to work with them, and why you need not.
Read the post
OpenClaw — The Engine Behind the Realm
What OpenClaw is, how the open-source AI assistant works, and why it is a solid foundation for running a crew of agents on your own machine.
Read the post
The Realm You Will Walk Into One Day
Zuckerberg imagined a metaverse full of strangers. I imagined something else: you log in and your entire team of AI agents is already there.
Read the post
Not All Agents Need the Same Brain
Running 9 AI agents on 4 different providers. Here is how REALM assigns the right model to each Character — and why one model for everything is a mistake.
Read the post
The Company of One Has a New Problem
Paul Jarvis wrote Company of One for a world without AI agents. That world is gone. What the company of one looks like now, and what breaks.
Read the post
Why Traditional Frameworks Break When Half Your Team Is AI
Scrum, Kanban, and old team models were built for humans. AI agents change coordination, memory, ownership, and the operating system itself.
Read the post