What the Machine Reads

Six months ago, I started building something because I couldn’t answer a question.

When an AI agent is asked to research a principal before a meeting — a potential investor, an advisory board candidate, a co-investment partner — what does it actually find? Not what Google returns when you search a name. What the model synthesises. What it weights. What impression it forms before any human in the process has opened a browser tab.

I had been using AI agents extensively in my own work — weekly CIO briefings, market monitoring, governance documentation for the family office. I had watched these tools become genuinely useful research partners. And I had noticed something I couldn’t unsee: before I engaged with anyone in a professional capacity, I was using AI to research them. So was everyone I was talking to. The question wasn’t whether this was happening. The question was what the machine was finding.

The build

What I built is called BrandStudio. It is not a social media monitor and not an SEO report. It starts from a written positioning record (the niche, the voice, three content pillars, and why the person is hard to substitute) and uses it to shape every public-facing asset, including a public bio page written to be read by people and AI assistants alike. Its free audit takes a LinkedIn headline and shows what is holding it back.

Building it meant looking hard at what the AI layer makes of real professionals across finance, technology and advisory services, starting with my own profile. What I kept seeing was a consistent and uncomfortable gap.

What I kept seeing

Professionals with decades of genuine experience, real credentials, and verifiable track records are, in many cases, either invisible to the AI layer or present in ways they would not recognise.

The machine is not reading your CV. It is not reading the biography your assistant updated three years ago. It is reading whatever the web has made legible about you — and synthesising that into an impression that travels faster than any introduction you have ever made.

In practice, this means two things. First, many experienced principals are effectively absent from the AI layer. The machine cannot find enough structured, coherent signal about them to form any useful impression. In a world where AI agents increasingly precede human due diligence, absence is not neutrality — it is a disadvantage that compounds quietly.

Second, where professionals do appear, the impression the machine forms often diverges significantly from how they understand their own positioning. The emphasis is wrong. The expertise is described in terms they no longer use. The work that matters most to them is exactly what the model cannot find.

Why this matters for principals

AI agents are already embedded in the workflows that precede professional relationships. Deal flow screening. Partnership evaluation. Advisory board assessment. Investor due diligence. The timing varies — but somewhere before a meeting is confirmed, someone in the process has asked an AI to tell them something about the person they are about to meet. That is not a prediction about where things are heading. It is a description of current practice.

What the machine finds in that moment shapes the meeting. It shapes the questions that get asked. It shapes the assumptions that enter the room before you do. The CIO who understands this is not playing a different game. They are playing the same game with information most of their peers have not yet processed.

The parallel to capital allocation is exact. A held position you have not reviewed is still a position. A professional profile you have not examined in three years is still a signal the machine is reading. The question is not whether the impression exists. The question is whether it is accurate, whether it reflects your current work, and whether it is doing the job you would want it to do.

The case against this argument

The obvious disconfirmer: warm introductions still drive most of the professional relationships that matter at this level. If the principals and CIOs who count are being introduced through trusted networks, the AI layer is noise — a pre-screening mechanism for people who were never going to make it through to a real conversation anyway.

That argument has merit. But it misses the sequence. A warm introduction gets you the meeting. What happens in the forty-eight hours between the introduction and the meeting — when the other party asks their AI to pull background on you — determines whether you walk in at parity or from a deficit. The referral opens the door. The machine decides what is waiting on the other side of it.

What to do

Treat your professional profile as a held position. Run it through the same analytical discipline you apply to the portfolio.

What does the AI find when it researches you? What does it weight most heavily — and is that the work you would most want to lead with? What is missing from the synthesised signal? What impression does it form in the first thirty seconds, before any human has made any decision about you?

These are not questions about personal branding. They are questions about whether the signal your professional reputation is transmitting into an AI-mediated environment is accurate, current, and working for you.

I also write The Principal’s Brief — a short note for family office principals and CIOs navigating the intersection of AI, commodities, and long-duration wealth. The issues are direct, practitioner-focused, and written to be read between meetings. If the questions this article raises are ones you are working through — what AI reads about you, what it means for your professional standing, how to position a family office in the age of intelligent machines — that is where I work through them in depth.

AI read your profile before the meeting was booked. Whether what it found represents you accurately is either a decision you made or a gap you left open.