One approach I've been thinking of experimenting with is saving everything I say, and using an LLM with my custom knowledge base to answer a question. I actually think a borg-like communication structure in a company could be really efficient. If every communication was publicly available, and we used ML to make that information consistent and to answer queries, I think you could achieve coherence with a company much faster, move as a unit much faster, and get a lot more done.
However, it's still a pipe dream at this point due to the context length limits. The actual amount of important context at a company is many times what a model can keep in mind now. As models get bigger, or if a group can run multiple models many times to make the company info widely consistent, it would be much better.
For instance imagine now, 30 people at your company work on one project, and 50 people work on another. They might take a slightly different definition for a term, and do work which is ultimately counterproductive to each other. Usually, it takes a conflict arising to resolve this, and people are long stuck with the mutually-incorrect definition. If an ML algorithm can look at everything both groups are saying and point out inefficiencies, I think you could get more cohesion, especially before problems get big enough to be a drain.
That being said, this type of system is more of a pipe dream, and I don't think it will be feasible for a few more years. In the meantime, I'm struggling how to balance communication with work. Advice appreciated.