We already have a successful application. But how can we use leverage AI to make it better ? This question gets asked often. While traditional developers are good at building web applications, generative AI is a new field that requires a different way of thinking.
Unlike the traditional systems, interface to generative ai models and APIs is different. You can not expect a deterministic answer anymore. The input variations can vastly change the output and this requires lot of testing before you can have the confidence of deploying it in production.
Our friends at Principle AI have been doing a great video series showing traditional developers how they can architect their systems to use LLMs and other generative AIs. This series is focused on seasoned engineers who are pivoting towards AI.
The idea to understand the basic building blocks of Generative AI and then taking few use cases around how they can be used to make your existing applications better. This will give both high level overview and low level details as the series proceeds.
The first video explains how to create an LLM infra at your company that can be used by others. This infra could just be a simple wrapper around the existing Gemini or OpenAI apis.