Blog

One Month with OpenClaw — A Non-Tech Person's Honest Take
I'm not a developer. I spent a month building an AI agent stack from scratch. Here's what actually happened.

Jensen Huang Confirmed It — This Is the New Computer
Jensen Huang says the agentic computer is the next paradigm. I've been building one for a month. Here's what the proof actually looks like from the inside.

Field Notes

The Operational Brain — One AI That Knows Your Business
Why the "get an agent" framing is wrong, how memory compounds, and what three years looks like.

Why I Tell Most Clients to Wait Before Automating Anything
Automating the wrong thing first is worse than doing nothing. Three questions I ask before recommending anything.

The Incidents That Made Me Build Guardrails I Should Have Built First
78 CMA PDFs with blank addresses. An AI responding to 20+ leads nobody asked it to contact. What happened and the four rules that came from it.

What It Actually Takes to Run AI for a Dozen Businesses
Version drift, silent failures, and why the economics only work with a fleet model. Field notes from one operator.

How Agent Memory Actually Compounds — and Why the First 90 Days Are the Hardest
Session memory dies. Structured memory compounds. What changes at 90 days vs. day one, and the discipline required.

The Platform I Had to Rebuild Mid-Flight
I built an agent platform, watched it become a liability, and migrated a dozen live clients without taking anyone offline.

My Terminal Is My AI Operations Center
Chat was my first instinct. The terminal is where I ended up. Fleet management, Claude Code, and four-command deploys.

AI & The Bigger Picture

Your AI Agent Doesn't Have a Model Problem
Everybody is asking which model to use. That is not the main question. The harness is the product.

The $300K/Month Agent: Why Felix Matters to Real Business Owners
An AI agent generated $300K in a month. Here's what actually happened and what it means for normal business owners.

How Local Businesses Get Recommended by AI
The practical GEO playbook for local businesses that want to be recommended by AI.

How to Get Recommended by ChatGPT
The GEO playbook for getting cited by AI assistants in local service queries.

Your Firm's ChatGPT Problem
Most RIA staff are already using ChatGPT for client work. The gap between consumer ChatGPT and the API is a live Gramm-Leach-Bliley exposure at your firm right now.

The BYOA Problem
BYOD was a hardware problem. BYOA is an intelligence problem. Top performers are building personal AI agents — and when they quit, all that institutional knowledge leaves with them.

The Business Case for AI Agents
ROI math, data sovereignty, and why local-first AI gives individuals more control than the SaaS alternatives.

Two Weeks, No Sleep, and What I Actually Built
Two weeks of all-nighters, one AI agent, and an operation that didn't exist a month ago.

Five Automation Workflows That Actually Save Time
High-ROI use-cases you can deploy without enterprise complexity.

Building the Stack Technical deep-dives on AI infrastructure, agent architecture, and lessons from production.

Nvidia Just Validated the Agent Infrastructure Category
NemoClaw is an enterprise signal for the whole agent infrastructure category — what it means now.

The AI Video Stack That Costs $0 Per Clip
What works now: local LTX experiments vs production cloud stack (Claude → Sora 2 → Topaz).

What Agent Architecture Actually Looks Like After 6 Months
Six months in production: what actually compounds (and what we'd do differently if starting over).

Obsidian as the Brain: Memory Architecture for AI Agent Systems
The practical blueprint for turning scattered notes into a usable memory layer for agents.

I Run 4 AI Agents From One Telegram Chat. Here's What Each One Does.
Role specialization, routing discipline, and real operational throughput.

Why Specialized AI Agents Outperform a Single Generalist
Generalist chat is easy. Operational throughput comes from role-specialized agents.

Apple Silicon Local AI Buyer's Guide
Every M4 tier from $599 to $3,999+ and the free models each one runs.

What I Learned Rebuilding an Agent Stack from Scratch
Model degradation, why local LLMs failed in production, and the reset that fixed it.

Memory Architecture for Agents
A 3-tier memory model that reduces drift and keeps context useful.

The Clean-Room Pattern: Cloud AI Without Leaking Data
Cloud AI performance without exposing sensitive information to third-party servers.

Why a $20 Model Beats a $200 Model (If You Do This)
The variable that drives agent performance isn't the model tier.

Your AI Is Only as Good as the Files You Feed It
Context hygiene beats model churn.

Fixing Agent Drift and Reliability
How to recover from silent regressions and restore execution trust.

Local vs Cloud Models for Automation
Where local LLMs fit, where cloud is safer, and how to run hybrid.

Agent Operating System Setup
Roles, guardrails, and proof-first execution standards.

Skills: The Loadout System for AI Agents
Why skills are the loadout layer that turns general models into dependable operators.

Making Agents Work While You Sleep
How asynchronous dispatch, checkpoints, and completion loops create overnight execution.

The Dispatch System
A practical model for dispatching work to on-demand agents without role drift or handoff confusion.

Sometimes I write about current events

Trump Doesn't Bomb on Mondays
5 major strikes. Both terms. 100% landed on Thursday or Friday. The pattern isn't coincidence — it's market awareness.

Who Owns the Texas Power Grid?
Interactive map: the companies controlling Texas power infrastructure are the same ones shaping its regulation. See who owns what.