Archive

Essays and posts

2026

September 2, 2026A Second Check, Alwaysessay
September 2, 2026Refuse Over Guessessay
July 30, 2026Kimi K3: The Architecture Was Decided By The Kernel
May 9, 2026Skills, Connectors, and Subagents: Decoding the Architecture Anthropic Just Made Standard
April 22, 2026The Open-Source Agentic AI: Kimi K2.6

2025

June 19, 2025Building Smarter AI Benchmarks

2024

March 20, 2024HITL vs Full Automation: A Decision Framework
March 15, 20245 RAG Failure Modes in Production

Notes

2026

September 13, 2026AI systems should earn autonomy through evidence, not confidence.
September 7, 2026Most AI agents remember. Very few actually learn.
September 2, 2026Use rank fusion, not a single score, to match messy records

Tags

agentic-ai (6)evaluation (6)model-analysis (3)notes (3)reliability (3)essays (2)architecture (1)automation (1)benchmarks (1)decision-framework (1)document-ai (1)hitl (1)inference (1)langfuse (1)learning-systems (1)llm-as-judge (1)open-source (1)production (1)rag (1)reasoning (1)reconciliation (1)regulated-industries (1)retrieval (1)verification (1)