Tag: ai
All notes with the tag "ai".
-
The Next Bottleneck Isn't Model Intelligence
The most interesting AI advances are shifting from bigger models to smarter systems around them — memory layers, grounding, inference efficiency, and modular intelligence.
-
HTML vs Markdown for Agentic Workflows: Why the Future May Be Hybrid
Markdown still dominates cognition-heavy AI workflows, but as agents become operational systems, HTML is re-emerging for structure, interaction, and state. The future is probably both.
-
When Fine-Tuning Actually Wins
Most companies shouldn't fine-tune first. But there are specific cases — cost, security, privacy, moat — where prompting and RAG quietly lose. Here is the honest map.
-
Small Models, Serious Work
Why smaller language models are quietly reshaping AI in 2026. The most interesting trend is not scale. It is restraint.