GAE as a platform is really great (good scalability, no maintenance overhead, low latency in most cases) but the extremely slow pace (compared to Amazon Web Services) for introducing new APIs and maturing the existing (ex.: increasing quotas and charging for the Search API, Prospective Search) makes me wonder: is Google really investing enough engineers to mature its platform? Will it pull the plug and stop offering it? Is the Premier support option really helpful? What are it's long-term plans? Where's the roadmap? These are very important questions considering it's PaaS and you'll be deeply committed to it "by design" once you Go Google.
I've been thinking about providing the opensource content-based image database at http://server.imgseek.net/ as a hosted service. This would allow image-related websites to provide visual query-by-example and query-by-sketch features with minimal setup effort. Can you see demand for it?
Visual search technology (content-based image and video query-by-example) has been applied to some problems, namely: copyright infringement tracking, image retrieval, mature content filtering, quality assurance etc. Is there any other application left to be explored by startups?
Considering that we already have some well established companies doing this (eg. http://picscout.com/, http://ideeinc.com) but aiming for huge clients (corbis, getty, jupiter), how big do you believe the demand for a free or low-cost service would be ? Think about the average professional or amateur photographer/image producer.