We are talking here about how to get your desired outcome. So, your focus is on this outcome and on the workflow architecture that leads to this desired outcome. The workflow is almost like a "lego" of tools, protocols and concepts which are (tools) created or dismissed (in the industry) as we speak. So, with everything so dynamic in the AI ecosystem, what do you do? There are a few rules on how to design the app that work and allow you some stability as everything else is changing. The rule number one is (ref. above) - you hold on architecture - all the guarantees of the good results are there. Keep both - very high level architecture and more detailed "classic". The rule number two is to ship the simplest apps possible because they might become absolute almost immediately. One more rule that works for us is use tools that are integrated with other popular tools, that makes them useful and propel to the front lines at least for a time being. (Ex. Hugging Face with Gradio, or Lovable with Github). Your old concepts, tools and systems, that you literally used yesterday or still are using, will not work where we are heading. This new world might require leaving behind almost all experience you have gained or things you have made and distributed before. Your current software stack, your established workflows, your entire mental model of how things get made. Abandoning the assumption that the way you've always created things (apps) is the way apps will always be created. (People have even spoken about the economic and social consequences of this change). What are the requirements than? - speed, open mind, knowing exactly a few fundamental design rules, being willing always simplify and DIV as never before. Will continue what i meant by all that. Let's design, share designs and deploy apps. Cheers inside.the