If you've worked with a coach for any length of time, you've noticed the same pattern. The session is sharp. You walk out with one or two genuine insights and a couple of small commitments. By Wednesday, the insight is fuzzier. By the following session, you can recall the vibe of what you talked about but not the specifics, and the coach gently reconstructs the thread because reconstruction is part of the work.

That's not a failure of you or the coach. It's a failure of the documentation layer between sessions. A coaching investment that holds up depends on whether the insight survives the week, and the cheapest way to make insight survive is to write it down in a place an agent can read across.

This isn't about replacing the coach. The coach does work an agent can't do — the relationship, the read on you, the asking of the right question at the right moment. AI notes earn their keep here when they hold the body of sessions and your weekly reality in one place, so the coach's question about "how did the experiment from last time land" has an actual answer instead of a guess. The hard-conversation recording loop here also shows up in [AI notes for co-founders: alignment, decisions, and accountability](/guides/founders-ceos/co-founders-alignment-decisions/) — same recording-with-speaker-labels mechanic, different relationship.

## A growth vault that holds the body of work

The shape that holds up is roughly: one top-level page for the coaching arc, with sub-pages per session. Each session page has the date, the recording, the transcript, your own notes from before and after, and the agreed commitments. One cross-cutting page per recurring theme — the pattern your coach keeps catching, the experiment you've been running for three months, the thing you said you wanted in the first session — that pulls forward the session moments where it came up. One inline database for active commitments with rows for commitment, date made, owner, status, and notes.

Capy supports unlimited page nesting, so a long arc with several themes can fan out into per-theme sub-pages. The commitments database lives directly inside the page via the `:::database:::` directive, so you don't switch tabs to a separate tracker.

The vault is plain markdown notes. That matters because the agent can read across every session, every theme, every commitment in one query. When you sit down to prep for the next session, you don't reconstruct from memory; you ask the agent to read the last three transcripts and tell you which threads are still open.

## Recording sessions with speaker labels

The single highest-leverage change you can make to a coaching practice is recording the sessions and keeping the transcripts. The reasons most people don't are real — it can feel like surveillance, it can change the dynamic, the coach might not be comfortable. Talk to your coach about it before you start; some are enthusiastic, some prefer audio-only, some prefer not at all. The mechanical capability is here either way.

If you do record, drop the audio on a Capy page. The transcript comes back with speaker diarization — labels like "Speaker 1: …" so you can tell who said what, not just a wall of text. Park the recording with the transcript on the session page.

The point isn't to re-listen to every session. It's that the next time you have a question — "what did the coach say about that pattern in the third session, was it as strong as I remember" — the answer is searchable. You can ask the agent to pull every passage across the arc where a specific theme came up. The session stops being a one-time event and becomes part of an evolving record.

## Reflection prompts that aren't a journaling app

The conventional advice on getting more out of coaching is to journal between sessions. The conventional reality is that most people stop within two weeks, because writing into an empty document with no structure feels like homework.

A working alternative: ask the agent to draft a reflection prompt for you each week based on the most recent session and your week's notes. "Last session you committed to noticing when you take on a meeting that should have been an email. Looking at this week's calendar and meeting notes, where did that pattern show up, and what did you choose?" The prompt is grounded in your actual week, not in a generic "what are you grateful for" template. You write a few sentences in response. The agent has the context for next week's prompt because your response is now in the vault.

Over months, this becomes a record of your own pattern recognition that the coach can read. It also becomes a record you can read back yourself, which is where most of the compounding value lives.

## Surfacing patterns across sessions

The work coaches do that's hardest to replicate is pattern recognition. They notice that you've used the same framing in three different sessions about three different topics. They notice that you say "yes, but" right before you reject an idea that sounds promising on paper. They've been holding the arc of you in their head and they're surfacing the patterns you can't see.

You can't replace that. You can give yourself a partial version of it by asking the agent to read across the transcripts. "Across the last six sessions, what topics have I returned to most often, what questions did the coach ask that I haven't answered, what commitments did I make and slip on more than once." The agent reads the transcripts and writes the answer. The patterns it surfaces aren't the same as the coach's, but they're a useful complement — and they're in your hands instead of the coach's.

Bring the patterns to the next session. The coach gets to react to what you've already noticed instead of having to find it cold.

## Integrating insight into the operating week

The reason most coaching insight decays is that it never makes contact with the operating week. The session ends, you go back to your calendar, and the insight stays at the level of "I should think about that more."

A working flow: at the end of each session, capture the one or two commitments in the commitments database with explicit owners and check-in dates. Then ask the agent to read the upcoming week and propose where the commitment most naturally lives. "You committed to running the meeting agenda by Joel before sending. Looking at next week, the standing Wednesday sync is the obvious place. Want me to draft a reminder that lands in your morning review on Tuesday?"

You're not building a productivity system on top of coaching. You're closing the loop between insight and the calendar where the insight has to actually land. The agent is the cheap glue between the session and the week.

## Reviewing the arc, not just the session

Once a quarter, do a heavier review. Ask the agent to read the last twelve sessions, the commitments you've made and slipped, the themes that have recurred, and write a one-page summary of where the arc has been and where it seems to be going. Bring the summary to the coach. Use it as the input to a re-contracting conversation: are the right things on the table, are the experiments worth continuing, what's actually changed. The same quarterly read pattern shows up at the company scale in [annual planning and goal setting](/guides/founders-ceos/annual-planning-goal-setting/).

This is the kind of review that's easy to skip and expensive to skip. Without it, coaching arcs drift, and you find yourself paying for sessions that are slightly less alive than they were six months ago. With the vault, the review is an hour of work instead of a weekend of reconstruction.

## What this isn't

Capy isn't a coaching platform. It doesn't have a coach-and-coachee shared view, doesn't broker the relationship, and doesn't generate session prompts that pretend to know your situation cold. The work happens in the room with the coach. Capy is the working layer for you between sessions.

It's also single-user by design. One person, one vault. Your coach maintains their own notes in whatever they use; you maintain yours in the vault. If your goal is a shared workspace where coach and coachee both edit the same notes with permissioned access, that isn't this product.

The vault is cloud-hosted. Your transcripts and reflections live on our servers, not on your machine. If the personal nature of the material makes you want a setup where the data stays on your own hardware, this isn't the right fit and we'd rather you know up front. For most people, a private cloud vault with one integrated agent is a strict upgrade over the current reality of session notes scattered across a notebook and a notes app on the phone.

## A small first test

If you've been coaching for a while, take the last three session recordings — even just your post-session voice memos if that's what you have — and load them on a single page in Capy. Ask the agent to read across them and write a one-page summary of the recurring themes, the commitments you've made, and the questions that haven't been answered yet. If the summary catches something you'd been carrying around without saying out loud, you've got a sense of what the agent does for the next twelve sessions. (For coaches working the other side of the table, [how coaches and trainers use AI notes to scale their practice](/guides/creatives-content/coaches-trainers-scale-practice/) covers the same arc from the practitioner's view.)

[Try Docapybara free](/accounts/signup/). Load three sessions and see what the arc looks like in writing.