February 14, 2026
Evals for Agentic AI: The Pillar That Makes Everything Else Work
Evals and data flywheels get lumped together constantly. I did it myself for a
evalsAI engineeringproduction AIagentic AILLM-as-judge
Read articleSLS Blog
Building AI You Can Trust
Lessons from the field on making AI systems honest, secure, and lasting.
January 11, 2026
Data Flywheels for Agentic AI: The Pillar Nobody Builds
At Sixfold, our feedback loop was a spreadsheet. A customer would flag an issue,
data flywheelAI engineeringproduction AIobservabilityevals
Read articleJanuary 8, 2026
Lessons from Building Real-World AI Applications
Three years of building production AI systems distilled into meta-rules and a four-pillar methodology. From training deep learning models to running an AI consultancy, these are the hard-won lessons about infrastructure, domain knowledge, evals, and data flywheels.
AI engineeringlessons learnedevalsproduction AImethodology
Read articleDecember 23, 2025
Five Things AI (LLMs) Excel At: A Bayesian Statistician's Perspective
A first-principles look at what LLMs actually do well, through the lens of Bayesian statistics and information theory. From lossy compression to posterior sampling, here's why certain tasks play to a transformer's strengths.
LLMsBayesianfirst principlesAI engineering
Read article