Bits and Bobs 2/16/2026

Idle agents are idle leverage

  • Idle agents are idle leverage
    • It feels like a wasted opportunity every moment your agents are not running.
    • If intelligence can operate asynchronously, then inactivity is lost upside.
    • What valuable work could be happening right now, even while I sleep?
    • How big is your team of agents? How many projects are they working on? Are you making the most of their time? Are they motivated to produce the best results?

Idle agents will turn us into even more of productivity monsters

  • Idle agents will turn us into even more of productivity monsters
    • Every industrial leap promised more leisure. Instead, it raised the baseline.
    • Factories did not shorten ambition. And as agents run 24/7, time becomes even more yield-bearing.
    • So AI may mean we'll have more time for "less mundane things", but markets will likely turn that into higher expectations from our work, as it always does.

The Four Levels of AI Aggregation

  • The Four Levels of AI Aggregation
    • From Ben Thompson and John Collison's discussion: AI commerce will not arrive all at once. It will stack.
    • 1 ) Do it for me
      You've already chosen.
      "Buy this."
      The agent executes.
    • 2) Help me decide (session-level context)
      You give context right now.
      "I need a winter jacket for 20° weather."
      The agent uses the constraints you just provided to narrow options. Every session starts fresh.
    • 3. Know me over time (persistent identity) (Shelf lives here)
      The system remembers what you like across sessions.
      Your past purchases, taste, habits shape future recommendations automatically. You don't restate yourself.
    • 4. Act ahead of me
      "It's October. You're moving north. Here are three jackets you'll love."
      The system anticipates needs based on accumulated context.

Self-Doubt is the real competitor

  • Self-Doubt is the real competitor
    • Most ideas die because of self-doubt. Because you didn't believe in yourself enough.
    • And once you start to doubt yourself, everything slows.
    • If you want more good things in the world, encourage the people building them — tell them you believe in them.
    • Conviction compounds.

People hear your idea through their own history

  • People hear your idea through their own history
    • What they've already tried. What disappointed them. What's currently broken in their life. What they're ready to believe.
    • There is no single clean explanation that works for everyone. The version that makes it click for you won't be the one that makes it click for me.
    • Here's the tension: The most valuable thing to build is a system. The easiest thing to understand is a tool.
    • Systems require patience and perspective. Tools are obvious and immediate.
    • We're trying to build systems in a world trained to look for tools.