AAIA - Episode 12 - What Should Your AI Actually Know
AI and Automation In Action
RAG Explained: How Retrieval Augmented Generation Makes AI Chatbots Secure, Specific, and Customizable
In this episode of AI and Automation In Action, Shane and Hunter explain Retrieval Augmented Generation (RAG) and how it scopes an AI chatbot’s “brain” to a defined set of business-specific context so responses come only from approved internal information. They contrast RAG with enabling web search or integrating external LLMs, and describe how workflows can route different question types to internal ticket data, a private LLM connected to a company website, or web-based sources for constantly updated information like compliance requirements. They also address security concerns, emphasizing that proprietary data stays protected only when both the chatbot and any connected LLM run as truly private instances. The conversation highlights how RAG enables granular control over sources, response formats, and outcomes, including custom API integrations that let chatbots take actions like retrieving or adding customer data.
00:00 Show Intro
00:35 Why RAG Matters
01:51 RAG vs Web Search
04:26 Mixing RAG and LLMs
06:33 Workflow Routing Examples
07:47 Keeping Data Private
09:44 Custom Actions via APIs
10:51 Wrap Up and Next Steps
11:16 Outro and Contact
Loading