Vocal for local
There was a Project I wanted to work on using both ChatGPT and Claude. My "elegant" solution has been to keep the files on my computer, and use Claude Code and Codex instead!
There is a new project that I’ve started on Claude which I expect to go on for a long time. I have a bunch of files there, which I keep adding to as I keep chatting with Claude about the matter (it generates these markdowns that keep getting added to the project).
Once in a while, I feel the need for a second opinion on some matter, and want to also talk about it with ChatGPT, where I have a parallel project on the same topic. After the first day of this project, though, things have gotten hard. I find myself needing to keep transferring files back and forth between ChatGPT and Claude.
This is unsustainable.
For a brief moment this morning, I wanted to look up for a solution that would allow me to share project files across ChatGPT and Claude. Before I could do that I realised an elegant solution exists - “going local”!
I created a folder on my computer and downloaded all the context files there. And henceforth whenever I want to work on this project, rather than using the Web or App interface, I’ll simply use Claude Code or Codex. This way I can even get them to check each other’s work, work on each other’s documentation, etc.
And thinking about it, I should just use this as the default mode of interaction for my work with LLMs where there are “attachments” involved. Rather than uploading the attachments to the LLMs’ clouds, I can simply get the LLMs to work on my computer and read and manipulate the files directly there.
In other words, I should just “go local”!
There is only one small problem with this - this restricts me to using all of these only on my computer. What if I have to look up something on the fly when I only have my phone? Maybe I should get back to finding that service that allows me to share projects across LLMs.
PS: In the middle of writing this blogpost, I tried Claude Dispatch. It is not very good, on first impressions.
PS2: One reason Dispatch is not great is because it sits within CoWork. And I don’t have a great opinion of CoWork as well. It is not particularly intelligent, wasting tokens instead on rather “fighter” solutions. Like this morning I thought I could use CoWork to download all my project files into one folder. It took half an hour and then booted me out for the session saying I’d “hit m

