December 7, 2025
Normies vs Power Users: FIGHT!
Context Plumbing (Interconnected)
AI that knows what you want before you ask — cool or just distracting
TLDR: Matt Webb pitches AI that acts on your intent by reading live context from your life. Comments split: a flashy mouse‑tracking UI derailed readers, and a debate erupted over whether regular folks can understand or control what their AI sees—key for trust as devices get more always‑on.
Matt Webb says he’s been doing “context plumbing,” wiring an AI so it knows what you want the instant you want it. He explains “intent” (your goal) plus “context” (what’s happening around you), and points to Do What I Mean and context engineering. Think wearables and always‑on cameras that watch, listen, and help before you even tap a screen.
The comments, though, turned chaotic. One reader fixated on the site’s mouse‑tracking effect: “cool, but distracting,” then admitted they stopped reading—peak irony for a post about removing friction. Others joked the AI should first “Do What My Mouse Means.” The sharpest take: users need an intuition about what context their AI can access. Cue a flame war: power users want knobs and dashboards; “normies,” as one put it, have used Google for decades without touching advanced operators. Privacy skeptics side‑eye lanyard cams, while hype fans dream of a concierge that just gets stuff done. The vibe: big promise, bigger questions. Can everyday people understand, trust, and steer an AI that’s peeking at emails, weather, and body language? Or will the “help” feel like surveillance with better branding?
Key Points
- •AI interfaces hinge on understanding user intent and providing relevant context to achieve goals.
- •Proximity to the moment and location of user intent drives product strategies toward wearables and embedded agents.
- •Agents perform better when prompts include rich, timely context; they operate via sequences of tool calls.
- •LangChain’s “context engineering” is defined as building dynamic systems that deliver the right information and tools for LLMs.
- •Context is dynamic and distributed across sources, requiring “plumbing” to collect, synchronize, and route it to AI systems.