Splitting text to manage context for agents and humans

Hey HN folks, I am building a new agent (desktop app) which uses text splitting as the core mechanism to manage context - both for humans and agents.

To allow users to have infinite chat without worrying about agent context, even sharing the chat with other users - say you want to mention legal team or dev team for a specific issue in your chat with agent.

Can use any agent/LLM internally. Would love to get your thoughts.