Managing Conversations

A conversation session is a continuous thread of messages between you and your AI employee on a particular topic or set of related tasks. Within a session, the AI remembers everything you have discussed and can reference earlier messages. Sessions persist until you explicitly start a new one with /new or until the context naturally resets. Think of sessions like phone calls — each one has its own context and purpose.

Use the /new command whenever you want to switch to a completely different topic. For example, if you have been discussing email outreach for a client and now want to research a competitor, starting a new session prevents the AI from mixing up the two topics. You do not need to start a new session for related follow-up questions or when iterating on the same task. The AI is smart enough to maintain context within a session.

The /branch command creates a fork of your current conversation. This is useful when you want to explore a different approach without losing your current progress. For example, if you are drafting a proposal and want to try a completely different angle, branch the conversation, try the new approach, and if it does not work out, you still have your original conversation intact. Branches are independent — changes in one do not affect the other.

Long conversations can sometimes exceed the AI's context window, which means older messages might get dropped. The /compress command addresses this by summarizing the conversation into a condensed form that preserves all key information while using fewer tokens. This is particularly useful during extended research sessions, complex multi-step tasks, or when you have been working on several related items in the same session. After compression, the AI still remembers the important details.

All conversations are stored in your dedicated database and can be referenced later. Your AI employee has access to previous conversations and can recall information you have shared in the past. The full Telegram chat history is also available in the Telegram app itself, searchable using Telegram's built-in search feature. For more structured recall, you can ask your AI employee to search its memory for specific topics or past discussions.