What happens in a private AI coaching session for an executive

How is a private AI coaching session designed?

I design every session around one piece of your real work, not around a curriculum. The structure follows the four steps on the coaching page: a discovery call, a map of your work, live build sessions, and then you deploy and iterate. Per session, that compresses into four moves.

1. Intake. You say what is eating your week and which documents or tasks are involved. We pick one.
2. Map. We walk through how the task is done today: inputs, steps, who touches it, where you lose time, where a mistake would hurt.
3. Build live. We build the assistant or routine together on your own material, run it on real input, read the output critically and fix what is wrong.
4. Hand off. You leave with something that works, a note on how to use it, and a next task for the following session.

Sessions are held on video. Length and cadence are agreed after the fit call, around your calendar. The goal of each one is at least one working tool or workflow you can use immediately. What we use depends on your stack: ChatGPT, Gemini and Claude are the usual starting points, and automation tools such as n8n or Zapier come in when a routine needs to run on its own.

What happens before the first session?

A confidential discovery call. We cover your biggest bottlenecks, the tools you already use and where AI could pay off first. I ask you to choose, before the session, what material you are comfortable putting into which tool. If your company has an approved AI account or a policy, we work inside it.

What are five illustrative examples of what could be built in a session?

These are examples of the kind of routine that fits one working session. Each is an illustrative example, not a client result. They are here to make the format concrete, and I make no claim that any particular person built or used them.

1. A document briefing routine

Illustrative example, not a client result. You drop in a long report. The routine returns a summary, the open questions, and the figures worth checking against the source. You stay the reader of record: it points you at the pages that matter.

2. An investor update drafting routine in your voice

Illustrative example, not a client result. You supply your notes and your last update. The routine produces a first draft in your phrasing. It drafts only. You edit and you send.

3. A meeting preparation routine

Illustrative example, not a client result. Given the calendar entry and your earlier notes, the routine produces a one page brief: who is attending, what was decided last time, what you want from this meeting.

4. An approval-gated inbox triage routine

Illustrative example, not a client result. The routine sorts incoming mail, proposes replies and flags what needs you. It never sends anything on its own. Every outgoing message waits for your approval.

5. A policy summary routine

Illustrative example, not a client result. It turns a dense policy or regulatory document into plain English and a list of questions to put to your counsel or your team. It is a reading aid, not advice.

Why do approval gates and careful handling matter?

Two reasons, both documented. The U.S. National Institute of Standards and Technology lists confabulation (confident but false output) and over-reliance on AI among the risks it identifies for generative AI, in its Generative AI Profile released July 26, 2024. That is why the routines above keep you as the decision maker. And in a 2025 global survey of more than 48,000 people, the University of Melbourne and KPMG found that about half of employees reported uploading company information into public AI tools. Microsoft and LinkedIn's 2024 Work Trend Index reported that 78 percent of AI users bring their own tools to work. Deciding up front which tool gets which material is part of the session, not an afterthought.

What is not done in a session?

• No implementation of enterprise systems. I do not deploy company-wide platforms, integrate your core systems or run a rollout inside a coaching session. That is a different piece of work, for your IT team or, where you need an operating role, a fractional chief AI officer engagement.
• No professional advice. A routine that summarizes a policy does not replace your lawyer. One that drafts an update does not replace your judgment.
• No audience. It is one to one. If you want to extend it to key people on your team, that is a separate conversation.

How is confidentiality handled?

Every engagement is conducted under strict confidentiality. I do not share client strategies, business details or AI implementations with anyone, and NDAs are available on request. The confidentiality page sets out the details.

What do you keep afterward?

Each session aims to end with at least one tool or workflow you can use the same day. You deploy what we built, and what support exists between sessions is agreed after the fit call. For the fee side, see what executive AI coaching costs and what the fee should cover.