Skip to main content
  • Journey
  • Projects
  • Publications
  • Tools
  • Blog
ESC
Start typing to search…
↑↓ Navigate ↵ Open ESC Close

Blog

Thoughts, notes, and write-ups on AI, engineering leadership, and system design.

  • May 9, 2026
    AICoding AgentOpinion

    Specs Are the New Lock-In (A Speculation)

    Speculation: AI coding companies may be moving the abstraction layer from code to language, and specs may be quietly trained to favor their home model. I have no inside information — just a hunch from my own usage.

  • Mar 8, 2026
    claude-codedevcontainersecurity

    Use .devcontainer as a Claude Code Sandbox

    A devcontainer gives Claude Code full autonomy inside a disposable environment, without the risk of --dangerously-skip-permission on your host.

  • Mar 1, 2026
    AIResearchProduct

    The Collapse of the Cost Barrier in the Research-to-Product Pipeline

    AI is collapsing the cost barrier between research insight and usable software, making bespoke, research-backed tools economically viable.

  • Feb 20, 2026
    OpinionCoding Agent

    Where Should a Coding Agent Live?

    AI coding agents are moving from synchronous IDE plugins to asynchronous infrastructure. Each placement changes what the agent can do and what it costs you.

  • Feb 20, 2026
    pythonreproducibilityenvironment

    Three Levels of Environment Control

    Reproducible environments prevent works-on-my-machine problems. The right level of control depends on project complexity — here are three practical tiers.

© 2026 Jason Lo
Email · Github (opens in new tab) · Linkedin (opens in new tab) · X (opens in new tab) · Scholar (opens in new tab) · ORCID (opens in new tab)