BLOG · CITED AI IN PRODUCTION

Notes on putting agents to work.

How investment teams move AI from demo to desk: process maps, agreed definitions, source trails, audit logs — and the portfolio graph underneath it all. Plus the Bayesian Investor primer for CIOs and selected work samples.

01 GRAFOLIO IN PRODUCTION

From pilot to desk.

Three essays on why desk agents stall — and the patterns that get them into production on real documents and Excel models.

Oct 2026 · agents

Why Investment-Desk AI Agents Stall in Pilot

Process maps, agreed definitions, source trails and audit logs decide whether a pilot reaches production — not the model.

Read →

Oct 2026 · excel

Excel Stays the System of Record

Agents should drive the approved workbook — not replace its math with AI arithmetic.

Read →

Oct 2026 · graph

Portfolio Graph, Not Folder Search

RAG over a data room finds documents. A portfolio needs entities and relationships you can traverse and cite.

Read →

02 THE BAYESIAN INVESTOR

A plain-language primer for the CIO.

How a team learns to think, model, and decide under uncertainty — one analogy at a time. No formulas, no Greek letters, no code. Start at the series index, then phases 1–4 and five desk use cases.

series index

Series Index: the Roadmap in One Page

The four-phase learning plan, the one-sentence takeaway, and a suggested path forward.

Start here →

the one idea

The One Idea Behind Everything

Start with a belief. Update it as evidence arrives. Always know how sure you are.

Read →

phase 1 · foundations

Phase 1 — Foundations

Probability as belief, updating a prior with evidence, Python as a what-if calculator.

Read →

phase 2 · modeling

Phase 2 — Core Modeling

Models, parameters, sampling, hierarchical models, regression — separating skill from market exposure.

Read →

phase 3 · tracking

Phase 3 — Dynamic Tracking

Kalman and particle filters for hidden, moving targets: the committee vs. the GPS.

Read →

phase 4 · integration

Phase 4 — Applied Integration

Sensor fusion, Black-Litterman, regime-switching networks: the full decision stack.

Read →

use case 1

Portfolio Construction

Who really moves with whom: hierarchical models for allocation under uncertainty.

Read →

use case 2

IPO Pricing & Feature Selection

Which factors actually matter: disciplined variable selection for pricing.

Read →

use case 3

Cross-Asset Stress Testing

When one domino falls: stress tests with private-asset look-through.

Read →

use case 4

Asset Allocation: the Operating System

Allocation as the system that turns beliefs and constraints into positions.

Read →

use case 5

Performance & Risk Attribution

The scorekeeper: Bayesian attribution that separates skill from luck.

Read →

03 SYSTEMATIC & ARTIFACTS

Models and work samples.

systematic · regime

Factor Rotation — Regime-Aware Systematic Portfolio

Two independent regime signals dynamically allocating across 9 uncorrelated factor ETFs: 2.6× SPY's Sharpe at under a third of its drawdown.

Read →

artifacts

Dashboard Samples & Artifacts

Screenshots from performance measurement, agent-based simulation, and multifamily operations work.

View →