Hi everyone,
I’d like to apply the Wisdom of the Crowd.
I am building a crew which surveys regulatory changes and tries to asses the impact on hardware components and operations. As depicted in the screenshot:
The idea is to survey the regulatory landscape, create a report summarizing it and then another research based on regulatory report to assess the impact on operations and hardware. For example, are specific pieces of hardware banned in certain regions.
I would like to know, from your experience if its better to build an agent all of the stages (using some for of RAG or memory) or delegating between two agents: a research one and a report creation one. The first will research regulation and pass on knowledge to the report creation agent who will return the report to the research agent who will analyze the impact on operations and hardware (according to findings from regulation report) and then the reporting agent will product a final impact assessment.
At the end of the day I would like to know the trade off between specialization and division of labor. Just saying currently using Gemini as the llm, but various models are an option if needed.
Thanks
