Tim, Alex, thanks for welcoming!
@Tim, I think we will be able to speak one language, as I was a product manager before. I also really want to see the usage of SingularityNet by commercial customers. This is the key to success for all of us, and this is the best motivator for developers.
Comparing to some seasoned professionals here, I cannot say that I am a mature expert in AI/ML. However, before joining SingularitNet it was my hobby to learn various stuff on this subject. When these activities were paid, the typical applications were:
- events prediction models (CART, Neural Networks, Regression, PCA, clustering etc.);
- identification of disorder in time series data (various EM algorithms, SSA).
Additionally I played around with OpenCV a little bit. At the moment I plan to work on OpenCog/AGI development.
Before answering your question regarding the market sectors, I think we need to identify the customers which we will focus on. You mentioned that we already have some completed services - I did not find anything about this. I will appreciate, if you share more details with me.
Presently, as I understood, SingularityNet is envisioned as a kind of M2M or partly B2B service, which should enable customers to create their own commercial services. There is a similarity to AWS/Azure/Google cloud based AI services in this regards. If all this is true, we might need to learn from stories of the customers of these service providers.
The USP of SingularityNet should be interoperability of its components and large ecosystem of agents. The later is a typical chicken-and-eggs problem, and I suppose that the present strategy to support creation of the first agents is right.
As for interoperability, I have an example. I worked with AWS and Azure. They have service implementing NLP algorithms to work with voice, however they do not have features to understand emotions, which is essential for medical services or CRMs. This might be one gap.
One more example. There are publicly available services, which processes frames and identify objects there. However these services are deaf and memory-less. Imagine we can provide a service, which can tell, is a person on another end of video conversation is currently lying. This might be an immediate success, especially in financial compliance domain.
Well, as you see, it might be a long talk. So Tim, once again, I would appreciate if you share your present understanding of the product with me.