4 September 2026
build in public on video, daily, without becoming an editor
a text update can be written by anyone, about anything, at any time, including afterwards. a video of you saying what broke today, filmed today, can only have been made by you, today. that is why build in public video works where text updates blur together: the face is the proof.
it fails for one reason, and it is not camera shyness. filming yourself is four jobs. decide what to say. set up. film it, usually three times. edit it. after a day of building, nobody does four jobs, so the channel gets one polished video a month and then none. the fix is not to lower the bar on the video. it is to take the jobs away.
deciding what to say is the job people underestimate. a blank camera produces a presentation, and a presentation is the thing that makes build in public video sound like marketing. an interview produces answers. underway asks five questions about the day you actually had, out loud, and you answer them. the opening question is written so the first thing you say names its subject, because an episode must never start mid-thought.
setting up is solved by not solving it. a laptop webcam or a phone propped on the desk, in the same spot, every night. nobody comes back for the lighting; they come back to find out what happened next. consistency of frame is a brand in itself, and one you cannot buy.
filming three times is solved by filming once. mistakes and pauses are fine. the cut removes the false starts and the filler and keeps what you said. a retake is for when you got the facts wrong, not for when you said them imperfectly.
editing is the job to delete entirely. underway cuts on full sentences so nothing is clipped mid-word, tidies long pauses, burns in captions, opens with a title card carrying your name and the episode number, and finishes with a small mark in the corner. it is ready in the morning. if a line should not be in it you drop the line. you cannot reorder last night into a better hook, and that is on purpose.
length is decided by the interview, never by the edit. three minutes of answers is the target and five is the ceiling, enforced by a clock rather than by the model's judgement. the cut removes filler and dead tangents and has no time budget of its own. an episode from a founder talking with real energy runs as long as they ran. cutting real content to hit a duration was tried, and it is the fastest way to lose someone's trust in the edit.
portrait is the default because a nine-by-sixteen file goes everywhere a short goes: tiktok, reels, shorts, snap. once a week a landscape sit-down, six to nine questions and seven minutes or more, is the long-form version for youtube and linkedin. same interview, more of it. the words around each are written for the platform they land on.
there is a version with no face in it. a travel day cut from the day's own footage with a voiceover recorded afterwards. an audio episode from a walk, with the questions kept in so it sounds like a conversation. both keep the number rising on the day the camera was not happening. neither is the default, because the face is the point.
the result is a serial, not a feed. episode 46 is watched by the people who saw episode 45 and want to know how it turned out. that audience is small at first and it does not leave, which is the opposite of a viral clip. the counter on the front of every episode is what carries it.
nothing here is faked. no reordering for a hook, no added words, no invented clips, no generated pictures. production quality is a separate axis and a high bar is wanted on it: clean cuts, clean audio, considered type. the honesty is about substance.
start tonight. the footage from today is the only footage of today there will ever be.