One morning. It is the only thing in the whole content plan that nobody and nothing can do for you — and everything else is capped by how well it goes.
5 August 2026 · read this on your phone while you set up
The production system is already built and tested
Forty-one tools are installed and working on your Mac right now, at zero cost: cutting, silence removal, vertical reframing for short-form, captions, colour, background removal, screen capture, and transcription. I tested transcription on a real file — 35 seconds of audio, 6.7 seconds to transcribe, word for word correct. All of it multiplies footage. None of it creates footage. That is what this page is for.
Why this comes first
The training footage is the ceiling on everything downstream, forever. Every platform — the avatar tools, the editors, whatever replaces them in six months — eats the same thing: good footage of you, shot consistently.
The footage is the investment. The platform is a rental. Shoot it once and properly, and switching tools later is a re-upload instead of a re-shoot.
What you already have. There is nothing to buy.
Kit
Verdict
Canon 5D
Fine. You're recording to the card and handing me the file, so none of the old questions about using a 5D as a live webcam apply here. Any 5D body works.
Blue Yeti Nano
Fine — and better than fine if the room is right. See below.
A tripod
The one thing you might not have. If you don't, that's the only purchase on this page and it's small — but it's your call, not mine. Books and a shelf genuinely work for one shoot.
The room matters more than the microphone
This is the cheapest quality win available and it costs nothing. A Yeti Nano in a soft, quiet room beats a far better microphone in a reflective one. Hard floors, bare walls, glass and tile are what make audio sound amateur — not the mic.
Pick the softest room in the house. Bed, curtains, rugs, a wardrobe of clothes nearby. A bedroom beats a kitchen every time.
Turn the aircon and any fan off while you're rolling. You'll hate this in Playa, which is exactly why the morning is the right window — before the house needs cooling.
Phones on silent, in another room. Not face-down. Another room.
Clap once and listen. If you hear a ring after it, the room is too live — add soft things or move.
The setup
Camera at your eye height, about an arm and a half away. Not looking up at you, not down.
Light in front of you, never behind. Face a window. Morning light, not direct midday sun — if you can see hard shadows on your face, soften it with a sheer curtain or move back.
Plain wall behind you. No logo, no branding, nothing with a name on it — the brand isn't picked yet, and anything branded dates this footage the day you pick one. A plain wall is reusable forever.
Yeti Nano just out of frame, roughly a hand-span from your mouth, pointed at you, on the front (cardioid) setting. Put it on something soft, not directly on a hard desk.
Record sound on both — the Yeti into the Mac, and the camera's own audio as a backup. I line them up automatically afterwards; it costs you nothing and it has saved more shoots than any other habit.
Do a 30-second test and send it to me before you commit to the whole session. I'll tell you if anything is off while you can still fix it.
What to record — three blocks, about an hour
Block 1 · about 5 minutesThe training footage
Straight to camera, natural speech, no script. Talk about anything you know cold — how the audit works, what you did last weekend. The words don't matter at all; this is the only block whose purpose is purely technical.
What does matter: consistent lighting, you looking down the lens, and normal movement. Don't sit rigid. This is the block every avatar platform trains on, and it's what makes re-testing a new one cheap later.
Block 2 · about 20 minutesThe founder story
The real one. What happened, what it cost, what you learned, why this business and why now.
This block never gets an AI version — permanently, not "for now." When people watch someone vouching for results, an AI presenter scores 43% lower on trust, and viewers can't say why it feels off. Your story and your proof are exactly that shape. They stay you.
Don't perform it. Say it the way you'd say it to one person across a table.
Block 3 · 30–40 minutesThe content bank
15–20 short takes, one idea each, 60–90 seconds. Say the point in the first sentence, then back it up. Stop. Reset. Next one. This is the block that gets atomised into everything else.
Keep them about the problem, not the industry
You haven't picked the vertical yet — restoration or foundation repair is still open. So talk about the problem, not the trade, and every take survives whichever way you go. Starting points, all of them straight out of the plan:
The three places the money hides: work already done that never got billed · revenue that never became a job · work people are being paid to do that a system should be doing.
Why the biggest one is money they've already earned — not new sales, not cost-cutting. Work performed, never collected.
Why you show clients the savings you refused to count, in red, struck through. And why that's the most persuasive thing in the whole pitch.
Why you don't get paid unless it works — and what that changes about the conversation.
What an owner thinking about selling should know: every dollar of profit you add is worth several dollars on the sale price.
The thing you got wrong in your own business first, and what it cost you to learn it.
Do a few of them twice, said differently. Not to pick the better one — so we can test which version performs, which is the whole loop you described: put a lot out, expect some to fail, learn, repeat.
What not to do
Don't read a script. It's visible instantly and it's the single most common reason this kind of footage looks cheap.
Don't change clothes, lighting or position between blocks. Consistency is exactly what makes the footage reusable — inconsistent footage can't be cut together later.
Don't shoot at your desk with your monitors and the room behind you. Plain wall.
Don't do it three times badly. One deliberate session beats three rushed ones, and re-shooting to fix the light costs you the whole morning again.
Then hand me the files
Drop the card contents and the Yeti recording anywhere on the Mac and tell me where. From there it's mine: transcribe it, cut the dead air, find the moments worth using, reframe them vertical, caption them, colour them, and give you back a first batch to look at.
Then we edit one by hand, together, and let what's annoying about it tell us what to automate. Building the pipeline before the first shoot would automate a workflow nobody has run.
Two things still waiting on you — and neither one blocks the shoot
Volume or precision. Your notes say hundreds of pieces a week across many handles. The plan's marketing section says the opposite — one specific teardown beats a year of posting. Both work; they can't both be true under one name. This decides where the content gets pointed, not whether you shoot.
Which platforms, and Spanish or not. Spanish is the one place an AI version of you genuinely earns its place — nobody's trust is on the line in a translated explainer, and nobody else in this market is doing it.
The shoot is safe under either answer, as long as the wall behind you is plain.