Blogs & Thoughts

Thoughts, career notes, and hard-earned lessons from 25 years+.

  1. Illustration: a navy shopfront with a Now Open sign, seen from the side. Behind the painted front is bare plywood, held up by wooden struts and sandbags like a film set.

    The Prototype Is Disposable. The Expectation Isn't.

    AI prototypes look finished, win the room, and set a bar the real system then has to clear. 'Build on it or start again?' is the wrong question. The code was never the expensive part. The promise was.
  2. Illustration: on the left, a man writes a script by hand at a desk by a window, surrounded by crossed-out drafts. The pages drift across to the right, where a machine with a film reel, a microphone and a ticked checklist turns them into videos on a desktop screen and a phone.

    The Content Engine: Automating Everything Except the Thinking

    I built a pipeline that turns a text file into a narrated, subtitled training video in my own voice, running entirely on my Mac. The code was the easy part. What it taught me about quality is the part worth sharing.
  3. Diagram of a local content pipeline: article and notes into OpenClaw, into a two-agent crew with voice documents attached, back out for review, then n8n

    Building a Local Content Crew with CrewAI

    A two-agent CrewAI pipeline that turns an article and my rough thoughts into a LinkedIn post in my voice, running entirely on local hardware. The build was straightforward. Teaching a local model to sound like me is the part I'm still working on.
  4. An S-curve diagram of two lines, technology and education, climbing through the Industrial and Digital Revolutions. Where the lines separate, zones of social pain; where they meet, prosperity. Adapted from Goldin and Katz.

    The Race Just Changed Gear

    The new pitch for agentic AI is that you no longer need to know how things work. Just state your intent and the machine handles the rest. The reality, buried in the same report selling that idea, is the opposite: the race between what you know and what the technology demands just found another gear, and the orchestrator everyone keeps promising does not exist yet.
  5. Diagram: OpenClaw inside an OrbStack container connected to Gemma 4 31B via Ollama, with Brave API for web access and a siloed Obsidian vault mount

    OpenClaw and Gemma 4: It's actually useful

    OpenClaw is a well-known self-hosted AI agent gateway with well-documented security concerns. Wrap it in a container, Tailscale it, and pair it with Google's multimodal Gemma 4 31B, and you've got something genuinely useful running entirely on your own hardware.
  6. image: Nvidia Slurm acquisition announcement with a diagram showing the pattern of open source projects being widely adopted, acquired, enterprise-tiered, and quietly deprecated

    Open Source Has a Landlord Problem

    Not all open source projects get acquired. But the ones that become critical infrastructure tend to follow a pattern so predictable it should have its own name. Widely adopted, acquired, enterprise-tiered, quietly deprecated.
  7. A split image: an ornate doorman on the left, an automatic sliding door on the right. One reads the room, one reads a sensor

    The Doorman Fallacy: Why AI Can't Replace What It Can't See

    A LinkedIn post about a doorman and an automatic door became one of the most-read things I've published. That reaction tells you everything about where we are with AI right now. People aren't looking for more efficiency. They're looking for permission to stay human.
  8. A Swiss Army Knife

    Generalists: The Hidden Asset You're Likely Overlooking

    The tech industry is obsessed with hiring narrow specialists. In an AI-driven world, this is a recipe for fragility. Here is why the generalist is your ultimate asset.
  9. A person looking at a completed maze from above, having never walked through it

    The Risk of Losing Hard Earned Knowledge to AI

    AI gives you the answer in seconds, but deprives you of the journey. Here is why the 'grind' of problem-solving is the most valuable part of your career
  10. A side-by-side of a university library and a busy server room

    Beyond the Degree: Why We Need a Better Proxy for Competence

    Earning a degree is a massive achievement. But when companies use it as the only filter for talent, they miss out on the battle-tested problem solvers they actually need.
  11. Weises's book cover

    The Cost of 'Ready Made' Talent: Lessons from 'Long Life Learning'

    Companies want pre-packaged, degree-holding experts. The reality is that surviving the AI era requires building an ecosystem of continuous learning.
  12. A clean architectural diagram of decoupled system nodes

    Understanding Message Queues & Event-Driven Systems

    A clear, instructional breakdown of how message queues work and why decoupling your architecture is the key to resilient systems.
  13. It's not on fire, but it will be fine

    Release on a Tuesday: Why Timing Matters in Software Deployments

    The logical time to deploy is when system traffic is lowest. The reality is that deploying on a weekend is a recipe for human burnout and fragile systems.
  14. A lone microphone on a dark stage

    Four Talks That Actually Change How You Lead

    Most leadership advice is corporate fluff. Here are four talks that skip the buzzwords and give you the foundational mental models to actually lead.