Founders: the writing you keep postponing

Last updated 11 October 2026

The specific problem

Founders do not have a typing problem. They have a fragmentation problem: the day is a sequence of conversations with gaps too short to start anything that needs concentration.

So writing accumulates. The investor update slips a week, then two. The reply to the customer who asked a hard question waits four days. The thinking about the hiring decision never gets written down at all, so it gets re-had in three separate conversations.

None of that is a discipline failure. Writing four paragraphs takes twenty uninterrupted minutes; you have eleven. Speaking four paragraphs takes ninety seconds, which you do have, several times a day.

What is worth dictating

The privacy point is sharper for founders

A founder's dictation is a remarkable concentration of sensitive material: term sheets, valuations, runway, salary bands, candid assessments of people, acquisition conversations, the things that are not going well.

With a cloud dictation service, all of that is transmitted to a third party in the ordinary course of using it. Nobody sets out to do that; it is what happens when you install a convenient tool without asking where the audio goes. On-device recognition removes the question.

The meetings are writing too

Much of a founder's week happens on calls, and the follow-up email is where the decision gets lost. Vaitly Meetings (beta, in Pro) records a call in the background the moment it starts on Zoom, Teams, Google Meet, Slack or WhatsApp. No meeting bot joins, and no Teams or Zoom admin permission is needed. The transcript is written on your computer; afterwards you get minutes, decisions and action items, filed in a calendar you can search.

Two honest caveats. You are responsible for telling the people on the call that you are transcribing it. And the summaries are written by sending the transcript's text, not the audio, to Vaitly's server, which uses OpenAI to write them.

A realistic day

1. 08:40, before the first call. Dictate three customer replies that have been sitting since Friday. Four minutes.
2. 11:15, between meetings. Dictate the hiring note from the interview that just ended, while the specifics are still there. Ninety seconds.
3. 14:00, after the board pre-call. Dictate the decision and the reasoning into the log. Two minutes.
4. 16:30. Dictate the investor update in one pass, loosely. Six minutes. Edit it in the evening, which now takes fifteen minutes rather than ninety.
5. Throughout. Dictate AI prompts that are sixty words rather than nine, and get answers worth reading.

Roughly fifteen minutes of speaking, replacing something like ninety minutes of typing that would otherwise have happened after nine at night, badly.

Setting it up

1. Teach it the names first Your investors, your team, your customers, your product. These are the words you will say most and the ones a general model knows least.
2. Make three snippets Your standard intro to a new contact, your "not right now" reply, and the standing context you attach to AI prompts.
3. Set a shortcut you can hit one-handed You will be using it while reading something else.
4. Start with replies, not documents The habit forms on short things. Long-form dictation is a second-week skill.
5. Accept the read-back Ten seconds, every time. Founder communications get forwarded.

Where it will not help

• Spreadsheets and models. Obviously.
• Anything you need to think slowly about. Speaking is fast, and some decisions benefit from the friction of typing.
• Open-plan or co-working spaces. Much of what a founder dictates should not be audible to the next desk.
• The pitch deck. Dictate the narrative, build the deck separately.
• Sending in anger. Dictation removes friction. Friction has saved a lot of relationships. Leave anything written at speed for an hour.