I participated in a competitive university hackathon recently. It was sponsored by some B2B cloud company peddling an "agentic platform" and an ill-defined (but also AI oriented) Google program.
Shockingly, all the cases demanded SaaS chatbot wrappers. I spent a good chunk of time commiserating with one of the other undergraduate teams about how cosmically lame this was, especially when contrasted with the "innovation" rhetoric espoused by the organizers.
To quote one girl: "Bro, my backend is just stateless Ollama calls and prompt templates. I could one-shot this garbage while hammered."
So, yes, they do feel fake. There's no challenge. The focus is on churning out box-checking slop rather than taking some kernel (a theme, perhaps, instead of precisely specified cases) and building something novel around it.
Shockingly, all the cases demanded SaaS chatbot wrappers. I spent a good chunk of time commiserating with one of the other undergraduate teams about how cosmically lame this was, especially when contrasted with the "innovation" rhetoric espoused by the organizers.
To quote one girl: "Bro, my backend is just stateless Ollama calls and prompt templates. I could one-shot this garbage while hammered."
So, yes, they do feel fake. There's no challenge. The focus is on churning out box-checking slop rather than taking some kernel (a theme, perhaps, instead of precisely specified cases) and building something novel around it.