My Personal AI Agent

For most of my career, I ran my information life the same way everyone else did. Alerts, tabs, newsletters, broker notes, forwarded emails — a constant inflow that I sorted, prioritised and acted on manually. As a data leader at Microsoft for eighteen years, I spent my days building systems that made organisations smarter. I came home and managed my own family office the way people managed family offices in 1995.

The irony took a while to land.

When I left corporate life and began managing the family’s capital full time, the gap became impossible to ignore. A well-run family office requires consistent research discipline, rigorous governance records, clear investment documentation, and the ability to synthesise information across multiple domains simultaneously — markets, regulation, tax, estate planning, and increasingly technology. A single person doing all of that manually is either working extraordinary hours or cutting corners they cannot see. Often both.

I was cutting corners I could not see.

What changed was not a particular product or announcement. It was a shift in how I thought about the technology I already had access to. I stopped using AI as a search engine — a tool I queried when I had a specific question — and started treating it as a persistent agent with deep context about my portfolio, my governance decisions, my investment thesis, and my thinking over time. The distinction matters more than it sounds.

What a Personal AI Agent Actually Is

The term gets misused constantly, so let me be precise about what I mean.

A personal AI agent is not a chatbot you query for information. It is not a productivity shortcut that writes emails faster. It is a context-rich working layer that sits between you and the complexity of your day — one that understands your domain, your decisions, your counterparties, and your outstanding questions well enough to do substantive cognitive work on your behalf.

For a family office CIO, that means something specific. It means an agent that knows the office’s current asset allocation, its investment thesis in enough detail to stress-test new opportunities against it, your governance calendar and outstanding items, your adviser relationships and what each one is responsible for, and the key questions you are currently working through. Trained on that context, a genuine agent does not retrieve information — it synthesises it in ways that are directly useful to the decisions in front of you.

Building that took time. It was not a single configuration decision. It was a deliberate process of structuring my own thinking in ways that an AI system could work with effectively — and then building a discipline around maintaining that context over time. That process became the foundation of both Vestry and BrandStudio, the two ventures I have built since leaving corporate life, which are essentially the inward-facing and outward-facing versions of the same infrastructure.

What It Does — Specifically

My personal AI agent now handles four categories of work that previously consumed between fifteen and twenty hours each week.

Research synthesis. Every week, I run a structured briefing that covers macro developments, commodity price movements across my specific exposures — gold, silver, uranium, copper, oil — APAC equity themes, and any regulatory news relevant to my holdings or structures. This used to take most of a Monday morning. It now takes twenty minutes and produces analysis that is better calibrated to my actual portfolio than anything I was producing manually, because the agent knows exactly what to look for and how to frame what it finds.

Governance documentation. The least visible and most consequential work in a family office is the paper trail — investment decision records, entity structure notes, adviser instructions, trustee resolutions. Most family offices do not maintain this consistently. I did not maintain it consistently. An AI agent removes the friction from this category almost entirely. When I make a decision, I record it in a structured format the agent can reference. When I need to reconstruct my thinking — for a tax review, an estate planning update, or simply because six months have passed — the record is there.

Scenario modelling. Quarterly, I run a structured scenario analysis across my portfolio: what happens under a commodity price correction, an Australian dollar appreciation, a rate cycle shift, or a specific geopolitical event relevant to my exposures. This is work that previously required either a dedicated analyst or a shortcut that substituted gut feel for rigour. Neither was acceptable. The agent makes it achievable for a single operator managing a portfolio of meaningful scale.

Content infrastructure. The fourth category is one that most CIOs and family principals have not yet considered seriously: using AI to build a consistent, high-quality external voice. The writing you are reading now is part of a deliberate programme — articles, newsletters, commentary — that positions the Pineda Family Office and its ventures in the conversation that matters to our counterparties and collaborators. That programme runs on roughly five to six hours of active work per week, producing output that would previously have required a content team. BrandStudio is the productised version of this infrastructure — what I built for myself, made available for other senior professionals who want to build a body of work without an agency budget.

The Two Layers

Vestry and BrandStudio are the same underlying insight expressed in two directions.

Vestry faces inward: governance documentation, family structure records, SMSF compliance workflows, estate planning co-ordination, and investment policy documentation — the operating system of a well-run family office. It is built on the observation that many APAC family principals in the $5 million to $50 million tier are running their most important financial affairs without written records that would survive a transition event, a dispute, or an ATO audit. The governance gap is not a knowledge gap. It is an infrastructure gap.

BrandStudio faces outward: content, positioning, and thought leadership for senior professionals who have spent decades building expertise and want to translate it into a body of work that compounds. The insight is the same one that built Vestry — most professionals in this category have no infrastructure, only effort. Content produced without infrastructure produces results proportional to the time invested, which means it scales poorly and stops when the person stops. Content produced with infrastructure compounds.

Both ventures are, at their core, personal AI agent infrastructure for a specific domain and a specific user — not software products looking for a use case, but infrastructure built backwards from lived operational problems.

What Changed

The clearest way I can describe the change is this: I stopped being reactive and became structural.

A reactive CIO responds to the market. A structural CIO builds a system that responds to the market on defined terms — with a thesis, a record, and a governance process that does not depend on the CIO having a good week. The difference in outcomes over a decade is not marginal.

The AI agent did not make me smarter. It made my existing thinking more consistent, more documented, and more actionable. That is a different kind of value — harder to see in a single week, unmistakeable over a year.

Where to Start

If you manage family capital — or any serious pool of assets — and you are still running your information life the way I was running mine three years ago, one change will produce the most immediate return: establish a structured weekly briefing discipline.

Write down your current portfolio exposures, your active investment thesis for each, and the three questions you are currently working through. Use that as the foundation for a standing weekly research session. Do not query for information — brief an agent with context and ask it to synthesise against what you already hold and what you are thinking about.

The quality of the output will surprise you. More importantly, the discipline of maintaining that context — of being rigorous enough about your own thinking to keep it current — will change how you make decisions.

That is the starting point. The next article in this series covers the specific tools and architecture behind the infrastructure — the stack I actually use, what I would change if I were starting today, and where the technology is heading in the next eighteen months.