The gear behind this blog: an 8 GB Mac, a keyboard and an old GPU

People ask what this site is built on, expecting a desk tour. The honest answer fits on three cards: a base M1 MacBook Air that writes, renders and deploys The Vibe Gate, the mechanical keyboard I type on, and an old gaming PC whose graphics card generates video locally. This page lists all three with their real specs, what each actually does, and where they fall short, so you can decide whether you need more than this before you spend anything.

What it does all day

Its job is production. Claude Code sessions run for hours on it, writing the articles' code and orchestrating the rest. Headless Chrome renders social graphics, and HTML compositions are captured frame by frame into vertical videos with ffmpeg. Four websites deploy from its terminal. I wrote up the numbers in The 8 GB Mac that runs four brands, including the one that surprised me: a 43-second vertical video renders in about 70 seconds.

Where 8 GB actually hurts

• Parallel renders time out. Two render workers at once can stall the capture, so long jobs run one worker at a time. Slower, but it finishes.
• Scheduled jobs are staggered. Automations that would overlap are spaced apart on purpose, because two heavy jobs at once push it into swap.
• No local language models. That's a job for a machine with far more memory.

None of that has stopped a project from shipping. It just means planning the order of work instead of running everything at once.

The keyboard: the one piece I reviewed on camera

Every word on this site is typed on it, and it is the only piece of hardware here I have put through a full review on camera, on my Arabic YouTube channel, which is embedded in The 8 GB Mac that runs four brands. The honest selling point is the sound and feel: lubed linear switches over a gasket mount, at a price tier where I did not expect it. Amazon lists it new at 4.6 stars from about 1,860 ratings as of October 20, 2026. If you touch a keyboard all day, this is the first upgrade I would make, before memory and before a GPU.

See the AULA F75 Pro on Amazon →

The second machine: an old RTX 2080 Ti PC

The Mac is the brain: it writes the prompts and assembles the final cuts. The PC is only a generation box. Cloud AI video generation is the genuinely expensive part of running content brands, so the 2080 Ti grinds out local generations instead. It is slower than the cloud, and its 2018 architecture lacks the newer number formats that modern cards use, so many models run slower on it than on a newer card. It was also paid for years ago, which makes it the cheapest line item in the operation. I have no buying link for it: it is an older build, and the card is long out of production.

If you want the same Mac

Apple doesn't sell this model new anymore. Amazon lists the same configuration (M1, 8 GB, 256 GB) as Amazon Renewed, meaning refurbished and resold, at 4.2 stars from about 2,580 ratings as of October 20, 2026. I haven't bought from that listing; it's here because it's the exact configuration this site runs on. Check the seller and the return terms on the listing before you buy.

See the M1 MacBook Air (Renewed) on Amazon →

What's deliberately not on this list

No microphone. The voiceovers in my videos are generated, not recorded; the setup is in Free voiceover with Gemini TTS and I auditioned three AI voices. No camera. The videos are rendered from HTML on the Mac or generated on the PC, not filmed. So if you're starting a similar project, a mic and a camera aren't the first purchase. Ship with what you have, and buy only the thing that's actually slowing you down.

As an Amazon Associate I earn from qualifying purchases. Ratings are from Amazon's listing on October 20, 2026 and change over time, which is why there are no prices on this page.