What Is an LLM? Models, Harnesses, Chatbots, Workflows and Agents | AI Training Session 1

What actually is a large language model, and why does the same model behave so differently in ChatGPT, a coding agent, or your own script? Session 1 of the Kinvecta AI training series builds the vocabulary everything else depends on. You'll learn: • What an LLM is (and isn't): next-token prediction, training, and frozen weights • Why one request often passes through several models • The rule of thumb: the model thinks, skills teach it how, tools let it act, and the harness decides which of those it gets • The six types of harness you'll meet at work, and how to size them up by reach and autonomy • Where the API comes from: vendor, cloud provider, or approved platform • The difference between a chatbot, a workflow and an agent: who picks each step? Includes a hands-on exercise. Pause the video to sort ten real tasks and inventory one AI tool your team uses. Chapters 0:00 Welcome 0:26 What you'll be able to do 0:52 Why vocabulary comes first 1:25 What an LLM is 2:13 How a model is made 2:47 Six beliefs to correct 3:29 One request, several models 4:04 The rule of thumb 4:23 The parts of an AI system 5:09 Same model, different harness 5:50 Six types of harness 6:39 Reach vs. autonomy 7:10 Three front doors for the API 7:48 Chatbot, workflow, agent 8:41 How far your environment lets you go 9:15 Exercise 10:03 Check your understanding 10:32 Next: AI resilience (Session 2) Slides (PDF), transcript and more for this session, free on GitHub: https://github.com/Kinvecta/ai-traini... The whole course: https://github.com/Kinvecta/ai-training Kinvecta: AI training, events and community. kinvecta.ai #AI #GenerativeAI #LLM #AIAgents #AITraining