That makes sense, but I'm not sure it's applicable to my use case, although I may be misunderstanding. When the user asks a question via raw input, which will be answered via the knowledge base, there are some instances where the best answer is in the form of an image. Eg "What areas do you service?" is best answered via an image of a map outlining the company's service area. This may not be possible but I'd like to find out 🙂 I suspect if it is possible it will need to be done via execute code but there may be a better way.