What we learned building it — Agentic AI, GenAI, ML, and DL, measured honestly, with a human in charge.
This is the AI Research Blog: what we learned building Agentic AI, GenAI, ML, and deep learning systems — measured honestly, with a human in charge. Each post explains the experiment, what it showed, and what it means for the next build. Nine posts, newest first.
- Machine Learning and Deep Learning: How the Models Learn — and How We Know They Work
- LLMs Explained: Frontier, Lite and Open-Source Models — and How to Choose the Right One for the Work
- Let SOC agents close the noise, not the attacks
- A Perfect Fraud Score Can Still Be Weak Evidence
- The Cheapest Setting Wins Only If It Still Gives the Right Answer
- “Positive” Isn’t Always Good News: Why AI Agents Need a Semantic Layer
- Your Prompt-Injection Filter Is a Smoke Alarm, Not a Lock
- Your RAG Tweak Might Be Noise: What a One-Change-at-a-Time Test Showed Us
- Why Your Domain Experts Should Teach Your AI
