Taking Generative AI Enterprise Models to Production

Generative AI (GenAI) is an emergent area that has many AI users excited. One key feature of GenAI applications includes using Retrieval Augmented Generation (RAG) to pull up text data from internal enterprise sources.   Another key feature is to reason about the retrieval results, the question, and the context of the enterprise LLM application. 

These features, RAG and Reasoning, have the advantage of reducing hallucinations and reducing the need for foundational model training on enterprise domain data (which may constantly be changing on a large scale). Reasoning improvements start with prompt engineering best practices, such as specifying the LLM application’s role, subject domain, type of detailed answer, format of answer, and providing multiple-shot question-answer examples.  Other reasoning improvements include Re-Act, to reason, and take action, to result in an observation.

SFbayACM hosts many lectures and training events. Please see https://sfbayacm.org for our upcoming full-day in-person workshops and monthly data science technical talks.