The blind spot in the AI meeting tool market
Search for an AI meeting assistant and you get a wall of products that do the same thing: join the call, transcribe it, summarise it, extract the action items. Granola, Otter, Fireflies, Fathom, Read — the list keeps growing, and every one of them is looking at the same forty-five minutes.
Nobody is looking at the fifteen minutes before.
That is where most of a meeting’s value is won or lost. The calendar notification fires, you glance at the name, and the questions start: what did we last talk about, did I owe them something, what is the context here. Then the call connects and you wing it.
That is not a transcription problem. It is a preparation problem, and a perfect transcript of a poorly prepared meeting is still a poorly prepared meeting.
What executive meeting preparation actually requires
A good chief of staff hands the principal a one-page briefing before every meeting. Not a LinkedIn summary or the company’s press page — your relationship with the person across the table. The checklist looks like this:
- Who they are to you. How you met, how often you talk, which projects or deals you share.
- The last interaction. Date, channel, the two or three points that mattered. Not a transcript — the distilled version.
- Open threads. What you owe them. What they owe you. Things discussed but never closed.
- Relationship health. How long since the last contact, and whether that gap is normal or the relationship is going cold.
- What to raise. Given all of the above, the one or two things this meeting should actually settle.
Run the checklist by hand and it costs five lines per meeting. At three meetings a day that is manageable. At five it gets thin. At eight, preparation time per meeting approaches zero and you are running on memory and hoping it holds.
The meeting preparation guide walks through the checklist in full. This page is about what happens when you stop doing it by hand.
Why walking into a meeting cold is expensive
It is not just awkward. You spend the first ten minutes re-establishing context. You ask questions that were already answered, and the other person feels unheard. You forget a commitment you made, and they notice — trust erodes slowly and invisibly. And you lose leverage: the person across the table came prepared, and you are operating at a disadvantage you cannot see. Senior executives in large organisations have staff for this. Founders have a calendar app and their own memory.
How PILOT prepares you
PILOT is a private AI chief of staff. It ingests your world — Microsoft 365 or Google mail and calendar, Teams chat and meeting transcripts, voice notes, messages — into a private knowledge graph that compounds daily. Because the graph already knows the people, the history, and the open commitments, a pre-meeting briefing is not a research task. It is a lookup.
Your morning briefing carries the day’s meetings with context. One briefing at 07:00, one big move, at most seven items, refreshed every three hours. Each meeting arrives with who, last contact, what is open, and what to raise — not “10:00 — Call with Thomas” but “10:00 — Thomas. Last spoke six weeks ago about the Q2 timeline. You said you would send the revised scope; it is still in draft.”
Ask on demand, in chat or by voice. “Brief me on Sarah.” “What is unresolved with the Hamburg project?” PILOT answers from the graph in seconds. The automated briefing is the baseline; on-demand questions go deeper.
Identity resolution does the matching. Meeting attendees are matched to people in your graph by email, alias, and context, not just by the name on the invite — so the briefing is about the right Thomas.
Nothing to configure per meeting. Connect your calendar once. PILOT watches the schedule continuously and prepares every meeting it sees, whether it was booked three weeks ago or ten minutes ago.
After the meeting: what was promised
A transcript-first tool ends when the summary lands. PILOT closes the loop. Teams transcripts, and Zoom recordings as an option, feed the knowledge graph after the call, so what was agreed becomes part of the record. Say what happened in a thirty-second voice note if you prefer: “Decided on option B. Sarah owns the plan. I owe Markus the revised numbers by Friday.”
From there, commitments are tracked in both directions — what you owe and what you are owed. They surface in the briefing as deadlines approach. A promise that stalls for ten days or more is flagged under “Pilot noticed”. Decisions go into the decisions log, with the context, so nobody re-argues them in three months. Next time you meet the same person, the briefing is sharper because the last meeting is already in it.
What this is not
PILOT does not put a bot in your call, and it does not send anything on your behalf — a human confirms every outbound write PILOT initiates. It is staff for one principal, not a team tool, with your data isolated by architecture in your own private database schema, EU-hosted on Azure. Built by its founder, it runs on his own calendar every day.
