ReflectionAI

Technologies like ChatGPT and GPT-4 start with a base large language model (LLM).  These models were trained using a vast amount of public data such as internet websites, books, articles, Wikipedia and public social media posts. 

When you interact with these AI ‘based’ models, they are able to generate responses, but the voice, content and personality captured by those responses are an amalgamation of the training data — meaning they lack a specific domain expertise or uniqueness.  By adding data, we’re able to make these models respond in a way that’s much more personalized. 

Personalized data can come in many forms – a collection of writings, articles, essays, even text messages and social media posts or your resume.