𝟭. 𝗪𝗵𝗲𝗿𝗲 𝗶𝘀 𝘁𝗵𝗲 𝗖𝗼𝗽𝗶𝗹𝗼𝘁 𝗴𝗼𝗶𝗻𝗴 𝘁𝗼 𝗯𝗲 𝗽𝗹𝗮𝗰𝗲𝗱?

Before starting your Copilot Studio project, consider where the Copilot is going to be placed, this is referred to as the channel and can have impact on things like adaptive card versions you can use!

𝟮. 𝗪𝗵𝗮𝘁 𝗱𝗮𝘁𝗮 𝘀𝗼𝘂𝗿𝗰𝗲𝘀 𝘄𝗶𝗹𝗹 𝘁𝗵𝗲 𝗖𝗼𝗽𝗶𝗹𝗼𝘁 𝗻𝗲𝗲𝗱 𝘁𝗼 𝗮𝗰𝗰𝗲𝘀𝘀 𝘁𝗼 𝗳𝘂𝗻𝗰𝘁𝗶𝗼𝗻 𝗲𝗳𝗳𝗲𝗰𝘁𝗶𝘃𝗲𝗹𝘆?

Identifying the data sources that your Copilot will use is crucial for its effectiveness. Determine whether your Copilot needs access to Dataverse, third-party products, informational files, or other business data. Consider the quality and format of these data sources, as well as any integration challenges that might arise (will you need to make a connection to the third party yourself?). Ensure that the data is current, well-maintained, and easily accessible to the Copilot, enabling it to provide timely and precise recommendations or actions. Don’t forget, you can upload the files directly to the ‘Knowledge’ section of Copilot Studio now, and it’s 512mb per-file!

𝟯. 𝗪𝗶𝗹𝗹 𝘁𝗵𝗲 𝗖𝗼𝗽𝗶𝗹𝗼𝘁 𝗹𝗲𝘃𝗲𝗿𝗮𝗴𝗲 𝗚𝗲𝗻𝗲𝗿𝗮𝘁𝗶𝘃𝗲 𝗔𝗜, 𝗮𝗻𝗱 𝗶𝗳 𝘀𝗼, 𝗵𝗼𝘄 𝘄𝗶𝗹𝗹 𝗶𝘁 𝗲𝗻𝗵𝗮𝗻𝗰𝗲 𝘂𝘀𝗲𝗿 𝗶𝗻𝘁𝗲𝗿𝗮𝗰𝘁𝗶𝗼𝗻𝘀?

Deciding whether to incorporate Generative AI into your Copilot is a critical step. Generative AI can drive more personalized, predictive, and context-aware interactions, you can now even use general-knowledge for generic responses. If you choose to use Generative AI, consider how it will enhance the Copilot’s capabilities, and if Generative AI is the best path. If you’re looking for structured data to be returned from a database, but also want dynamic generated information from files to be queried, maybe you could consider both a Generative AI and a non-Generative-AI topic approach.