"Generative AI is easy!" you ask a question, you get an answer! How difficult could it be? Well, for you as an individual, it is easy and simple to use, but for an enterprise to allow the use of Generative AI, it is very involved to make sure all the pieces are secure, scalable, and extensible. There are servers, clusters, APIs, Agents, Tools, Datasources, Prompts, Orchestrators, Embedding Models, Vectorization Databases, and many other endpoints that must be put in place and secured to provide reliable answers to questions using Generative AI in a private and secure setting for an organization.

The goal of this session is to go through all the different pieces that need to be in place for an enterprise to pass the scalability and security test while maintaining compliance in its own industry whether Fintech, Healthcare or others...

Remember that having 10 people using a Copilot is different than having 25,000 people using it. Also remember that vectorizing 10 documents is different than vectorizing 250,000 documents and most likely traditional RAG will fail at that scale. We will discuss Knowledge Graphs and what they bring to the table to exceed the traditional RAG approach. We will also discuss building solutions that require multiple LLMs (OpenAI, Gemini, Grok, Claude, Llama, Phi and others...) with multiple orchestrators (Langchain, Semantic Kernel) and multiple datasources (SQL DBs, Documents, APIs) and multiple Vector Stores (Azure AI Search, Azure Cosmos DB, Azure PostgreSQL, Pinecone, others...)