When Generative AI such as ChatGPT was launched it was a conversational style chatbot but now has evolved into an AI assistant carrying out multitude of functions. Prompt engineering was one of the first practical skills to emerge around generative AI. Once people began using tools such as ChatGPT, they quickly noticed that the wording of a request mattered. A carefully worded prompt could tell a model what to do, yet it could not give the model access to a firm’s internal documents, recall every relevant part of an earlier conversation, or retrieve information that had changed since the model was trained.
This limitation leads us to the solution which we now call as “Context Engineering”. Think of it as a full set of information placed in front of the model for it to access before starting the task or during the task. Prompt engineering remains part of the process, but it is no longer the whole process.
This post is published in full on my Substack.