The current generative AI journey for any product
Databricks provides a good model for approaching the product journey for gen AI products. It goes like this:
Start out with prompt engineering and build a finished product just using prompt engineering
Advance to RAG, providing more context to your models
Then, start to fine-tune the models
Finally, pre-train your own models (largely)
The great thing about generative AI is that you can provide a good user experience at every step! The more you advance across those steps, the more complexity you add, and the more you build a competitive advantage. But your users get value from day 1.
