A.I. is starting to be part of nearly everything people do with software. I see there are a lot of recommendation engines out there, but which ones are helpful and free or reasonable for startups? What's your experience been as a CTO or cofounder in terms of adding APIs from recommendation engines? Google's Prediction Engine is deprecated and the NLP API is amazing, but seems to have little value in terms of recommendation. What really works in your own experience?
Truth is, with AI being such a key differentiator for companies of all size in 2018, machine learning initiatives should start from the inside.
> Universal APIs are great for solving simple problems that almost every single company faces, such as translation or text sentiment analysis. There's tons of simplistic recommendation engines out there working well as long as all you have in terms of data is user IDs and product ratings.
However the moment you need to go beyond basic cases and leverage your company's big data in a meaningful and optimal way, the only path is to build a tailor-made architecture with an experienced data science team. If you allow me to use an analogy: the same dilemma exists in car racing, one could join a racing competition with a publicly available sports car, but at the end the winning team will be the one who built a custom f1 car.
If this topic interests you, I covered the major pitfalls leaders make with machine learning in a recent article: Medium article