The Fundamentals of AI Engineering - EP 01

AI Engineering is the process of enhancing real software applications with useful AI features using existing AI models. This involves connecting a suitable model to the application and ensuring the feature is reliable. Laravel developers are responsible for connecting the model and making the feature work correctly, while the AI model handles understanding information and suggesting answers. The Laravel framework manages permissions, validation, database operations, and other business decisions.

In an e-commerce application, AI engineering might involve recommending products based on browsing behavior. The AI model is only one component of this feature, with other components including selecting a suitable model, providing relevant data, designing prompts, retrieving information from a knowledge base, and handling errors and failures.

Traditional Laravel features follow a deterministic process, while AI-assisted features can evaluate signals that are harder to represent as fixed rules. For example, an AI-powered recommendation might consider recently viewed products, previous purchases, budget, stated preferences, and product compatibility.

To understand AI application architecture, it's important to distinguish between four concepts: LLM (Large Language Model), provider, model, and prompt. An LLM is a type of AI system trained to understand and generate language, a provider hosts AI models and exposes them through an API, a model is the specific AI model used for processing requests, and a prompt is the instruction and input sent to the model.

The Laravel AI ecosystem goes beyond just an HTTP client for a model API, including integration layers, providers, agents, tools, schemas, retrieval systems, protocols, and queues.