Home

ZiptieAI demos/documents AI (primarily LLMs), from “tiny” neural networks to enterprise AI (Palantir).

• AI fast track
  • Hack explains the core of what makes AI tick.
  • QS covers selected demos.
  • Concepts summarizes the core concepts (mainly from the QS demos).
    
    
• AI deep dive
  • 2 NNs demos NNs, CNNs, and transformers (TFs) (the source of the “I” in AI). In this section you control the AI mathematics.
  • 2b Models demos how to build packaged AI models (primarily LLMs). Covers fine-tuning, local models, deployment, and model serving (I am convinced that local models, like local computers 40 years ago, will become the norm). In this section you control the model.
  • 3 Agents. Python agents that implement reliable workflows, using AI models as helpful assistants. You control the application.
  • 3b Workflow platform (such as n8n) demos of practical agentic AI applications. The workflow platform controls the architecture.
  • 3c Enterprise AI demos (current focus is on Palantir). The enterprise platform controls almost everything.
    
    

ZiptieAI focuses on providing insightful and practical guidance with minimal text and diagrams (not on the Chicago comma). If I was working on a site for a customer, I would focus on customer reuqirements (even if that included the Chicago comma).

The following diagrams summarize the core ZiptieAI take on AI.

AI = deterministic computational algorithms that are extremely useful for specific types of applications

AI intelligence is a fairy tale ((internal) agent + TF (transformer) = LLM)