Also, if somebody has an idea for the following: we use spaces as client spaces. Meaning as a consulting firm, we have spaces for some clients, that includes all the relevant data. We built agents that execute specific jobs (e.g. analysing from a marketing and sales operations perspective, SEO etc). I understand an agent can operate within or access only one space at a time. This means, we have to replicate the "horizontal" agent (e.g marketing_agent) and configure so that the marketing_agent can access client A, name it as marketing_agent_clientA, then do the same for client B, and hence have marketing_agent_clientB.
If I change the prompt / logic in marketing_agent, these changes do not automatically transfer to the other client focused replica. I would have to go into each client agent and apply the changes manually. Is this correct?
One way around this would obviously be to allow agents to invoke a space through the prompt or within the conversation, so that instead of the agent itself only the conversation builds thematically around the client. This capability does not exist, as far as I know.
Any thoughts?
