Dang…
Your apps, your stores, whatever you’re building — your interface is now Claude, ChatGPT, Gemini or OpenClaw. And I guess you’ll soon be able to farm your compute across your devices. That might not be readily intelligible to most people, so, concretely:
Three computers, each building multiple projects, while OpenClaw is building on “Billy in Philly,” the Mac mini. Build more before my 19-month-old kicks me out of my office for bed.
Part of the “soon” arrived today. OpenAI put Codex into the ChatGPT app on iOS and Android, so that “from your phone, you can work across all of your threads, review outputs, approve commands, change models, or start something new.” OpenAI’s own framing, in the same report, was that “This is more than the ability to remotely control a single task or dispatch new tasks to your computer.” Alongside it, SiliconANGLE reports, comes Remote SSH, which “enables Codex to connect to remote development environments via an encrypted network link,” and Hooks, for customizing the agent with scripts.
That is roughly the shape of what I’ve got running, assembled by hand. The bottleneck is the one you’d expect. Memory. Each project holds its working context in memory while it builds, and when the machine runs out, a project forgets where it was. A machine building several at once starts dropping the thread on some of them.
So I actually think I need one DigitalOcean droplet for every project — dedicated to each, so I can have each one up in a browser tab and none of them losing progress to memory constraints. A droplet is DigitalOcean’s name for “a Linux-based virtual machine (VM) that runs on top of virtualized hardware,” and the smallest starts at $4 a month for 512 MiB of memory and one vCPU. Bigger ones cost more, but a dedicated machine per project is a small line next to the model subscriptions.
Or, ChatGPT’s distributed compute.
