AI & Technology
The Access Cliff: The Best AI Model Isn’t Buildable
The best AI model and the model you can actually build on are drifting apart. Here is the access cliff, and why picking the smartest model is the real trap.
AI & Technology
Essays & insights on ai & technology.
AI & Technology
The Authenticity Premium: Why Proof Beats Output
When AI makes output free, buyers stop paying for the thing and start paying for proof it is real. The authenticity premium is where value moves next.
AI & Technology
The Verification Ceiling: Why Cheap AI Slows Teams
Cheap AI made generation nearly free, but verification is the new bottleneck. Why checking AI output caps your team's throughput, and three ways to lift it.
AI & Technology
The Capability Overhang: Why AI Value Has Moved
The capability overhang is the widening gap between what AI can do and what companies actually use. Here is why it grows and where the real value sits.
AI & Technology
The Governance Gap: Why AI Agents Stall Before Production
Most enterprise AI agents die between demo and production. Not on model quality, on the governance gap: who is accountable, can you audit it, can you stop it.
AI & Technology
Context Rot: Why Bigger Windows Make AI Worse
Context rot means a model's effective context window is far smaller than the number on the box. Here is why bigger windows make AI worse, and how to feed less.
AI & Technology
The Inference Shift: Where AI’s Real Cost Moved
The inference shift moved AI's real cost from training models to running them. Where the money, the constraint, and the founder opportunity actually live now.
AI & Technology
Why AI Agents Ship in Regulated Work First
AI agents reach production in regulated, auditable work first. The moat is the audit trail, not the model. Here is the deployment order and why.
AI & Technology
The Evaluation Gap: Why AI Agents Fail in Production
AI agent evaluation is where demos lie. Why agents pass benchmarks yet fail in production, and how to measure real production readiness before you ship.
AI & Technology
Computer-Use Agents: The Interface Ladder
Computer-use agents can automate any software a human can see, but they are slow, costly, and brittle. The Interface Ladder shows founders when to use them.