Chances are you should instead be investing in data engineering so you can take a deep look at your properly labeled and processed data instead of throwing hundreds of thousands of dollars in Machine Learning (ML).
ML isn’t magical, you need a lot of clean, labeled data. Then you need a specific use case to design and deploy a great model that can help your product. Most startups don’t have clean, reliable and labeled data that can be used to train models. That’s why you need data engineers.
The reality is that given a specific opportunity and the right data, your product & engineering team should be able to build a decently fast solution and get you to 80-90% there. ML is awesome at giving you the last 10-20% and to not require an entire eng team. But building, validating and deploying a ML model is time consuming. Plus, very few people have experience going through this process which means higher risks.
My recommendation: 1. hire data engineers and build an amazing data set 2. test your hypotheses quickly and in prod 3. invest in ML to replace the validated “80% prototype”
I originally posted this note on Linkedin, see insights and comments there