Every Company Can Have a Quant: A Decision Algorithm That Keeps Improving
Llama Research runs a decision algorithm that forward-tests from a flat baseline, adds only variables that work, and recalibrates when it fails. That loop is a firm's quant — on the S&P 500 first, then on proprietary client data.
The Trials of Building a Quant Trading Agent — And Why We Still Won’t Let the Model Drive
A plain-English look at creating a quant trading agent: a dead ranker, three signals in test, and why we sit still when the model has nothing to say.
Building Auditable AI: How to Design Agent Systems That Can Explain Every Decision to a Regulator
Learn how to design AI agent systems with built-in auditability — decision logging, reasoning capture, confidence scoring, and human-readable audit trails for regulators.
Why This Discipline Matters Even More for Your Own AI Agents
The dangerous AI agent isn't the one that crashes — it's the one that looks like it's working. Five disciplines that make an agent team safe to delegate to.
Why Most AI Projects Quietly Fail — and the Simple Discipline That Fixes It
Why most AI projects fail: the model is graded on a proxy, not the outcome. Five validation rules — second opinion, forward testing, decay monitoring, and a human sign-off before launch.
Private AI That Survives an Audit: 7 Controls CISOs Actually Sign Off On
Seven private AI controls CISOs sign in audits: egress, classification, model inventory, identity, decision logs, HITL gates, and AI incident response.
What Human-in-the-Loop AI Actually Costs — and What Your Board Should Be Tracking
Human-in-the-loop AI cost and governance: latency, review load, oversight infrastructure, and 5 board metrics that prove HITL is working — not theater.
How Project Cortex Decides Buys and Sells: Multi-Agent Trading Decisions, OpenClaw Orchestration, and Obsidian Memory
How Project Cortex decides buys and sells: multi-agent trading decisions with weather → neighborhood → name → rails, OpenClaw orchestration, Obsidian memory, and a human gate.
From Research Platform to AI Factory: How a Decision System Learns From Its Own Work
What an AI factory means for decision systems (not GPU marketing): research platform → governed factory with propose-deploy rails, QA gates, and lessons that compound.
Human in the Loop AI Is Mostly Theater — Here's What the Real Thing Looks Like
Human in the loop AI is often theater. Real HITL / human-in-the-loop agent design: propose vs deploy, named veto, immutable logs, and fail-closed behavior — how to test yours this week.
