Beaker sends messages to be summarized in a format that cannot be viewed by the agent, so the agent can only summarize based on context in other messages.
Passing multi-modal messages as multi-modal would allow the agent to summarize based on the image contents rather than just context.
Beaker sends messages to be summarized in a format that cannot be viewed by the agent, so the agent can only summarize based on context in other messages.
Passing multi-modal messages as multi-modal would allow the agent to summarize based on the image contents rather than just context.