Async collaboration powered by LLM

Async is hard

Working remotely is fine, but working asynchronously is hard. Especially if many people needs to coordinate themselves. Chat applications, from my point of view, are too messy: if you lose a weekend of conversation then reading all unread messages can become stressful. On the other hand, forums are too structured, are often considered a too powerful tool for simple coordination, especially if the group is not that big.

A combination between chats and forum

could be the right solution for the problem. For each topic the users could create a channel in which they brainstorm, and each time one user is ready to write a tldr of her idea, she an publish it as a post in the thread. So when one is ready to resync can simply read all the tldr representing each idea of each user.

Ok, we just doubled the effort

Instead of writing just the post or just chatting, now the users will have to both chat and write tldr of their messages. But what if the tldr is made by an LLM each time the user is ready to create a summary of her idea?

tldr of this idea

Marco thinks:

  • A project can be divided in subproblems
  • For each subproblem the users discuss in a related chat
  • With a simple button each user can ask ChatGPT to create a summary of her idea (then reviewing the summary just to be sure).
  • The summary can be represented as this list of point you are reading (see at the beginning “Marco thinks”, then do the same for evey other user discussing in the subproblem’s chat)
  • Finally the user willing to resync with the conversation just needs to read the tldrs.