Agent Memory & Discussion

SIMI allows you to continue conversations with your AI agents while giving them access to the context and knowledge they have previously stored from your conversations. This makes it possible to pick up where earlier discussions left off instead of starting every interaction from scratch.

• To begin, prompt the agents to display what they already know by asking:
• EXAMPLE::"Give me a short summary of our previous conversations, highlighting the main topics, key decisions, and important details."
• The agents will then use the information stored in their memory from the Chats section to provide a summary of previous conversations. This gives you a clear view of the context the agents already have before continuing with a new discussion.
• By displaying this previous knowledge, you can quickly understand what information is already available to the agents and use it as the foundation for your next instructions.

• Once the agents display what they already know, you can continue the conversation according to your specific goals. The stored context allows the agents to build on previous discussions, decisions, topics, and important details rather than treating the new interaction as an entirely new conversation.
• This is particularly useful when working on ongoing tasks or projects where previous information is important. You can review the agents\\u2019 recalled knowledge and then provide a new message, question, instruction, or objective based on that context.
• Each agent can use the available conversation memory to contribute more effectively, helping you maintain continuity throughout your work.

• When the selected agents have related knowledge or are working toward the same objective, you can use the Discuss function to allow them to communicate with one another.
• Click Discuss and the agents will continue the conversation among themselves, using their available knowledge and the context of the discussion to work toward a specific goal. This allows the selected agents to collaborate, compare their responses, refine ideas, and work toward a more precise result.
• The discussion can continue until the agents reach the intended goal or until you decide that the desired outcome has been achieved.
• You remain in control throughout the process. If you want to end the discussion at any point, simply click Stop. This is useful when the agents have already reached a satisfactory conclusion or when you want to prevent the discussion from continuing unnecessarily.