Model Context Protocol (MCP)
A protocol for connecting AI systems with external tools and data sources, extending what a model can see and do.
Agents & ReasoningRetrieval & Context
Glossary
AI Concepts
A growing glossary of AI concepts and terminology — the building blocks behind LLMs, retrieval, agents, and reasoning that every AI PM should know.
AI Agents: What They Are, How They Work & Examples
A longer-running AI process that uses models, tools, and other systems to take a sequence of actions toward a goal.
Agents & Reasoning
Reasoning Models
Models designed to solve problems through additional inference — such as chain-of-thought — rather than an immediate answer.
Agents & Reasoning
Multimodal AI
Models that can work with multiple types of data — text, images, video, and audio — instead of just text.
Fundamentals
Context Summarization
Compressing older conversation history into a shorter representation to control context size and inference cost.
Retrieval & Context
Context Engineering
The broader practice of assembling and managing everything an AI model needs to produce a useful result — beyond a single prompt.
Retrieval & Context
Retrieval-Augmented Generation (RAG)
Combining information retrieval with language generation so an LLM can answer using domain-specific context.
Retrieval & Context
Vector Database
A database that stores vector representations and enables fast similarity search — the retrieval layer behind most RAG systems.
Retrieval & Context
Small Language Models (SLMs)
Smaller models built with fewer parameters and resources than an LLM — cheaper, faster, and often good enough.
Deployment & Efficiency
Quantization
Reducing the numerical precision of a trained model's weights to cut memory and compute needs at inference time.
Deployment & Efficiency
Knowledge Distillation
Transferring useful behavior from a larger 'teacher' model into a smaller, cheaper 'student' model.
Deployment & EfficiencyTraining
Reinforcement Learning
Training a system using feedback about the quality of its actions or outputs, reinforcing what works.
Training
Fine-Tuning
Adapting a pre-trained base model to perform better for a specific task, domain, or response style.
Training
Few-Shot Prompting
Providing examples inside a prompt so a model understands the expected task or response format — no retraining required.
Retrieval & Context
Self-Supervised Learning
A training approach where the structure of existing data provides the learning signal, without manual labeling.
FundamentalsTraining
Large Language Model (LLM)
A neural network trained to predict the next token in a sequence — the foundation of modern generative AI.
Fundamentals
Transformer
The model architecture — built on attention and feedforward layers — behind modern language models.
Fundamentals
Attention
The mechanism that lets a language model use surrounding context to interpret a word or token correctly.
Fundamentals
Vectors
A numerical representation of meaning in multidimensional space — the basis for semantic search and retrieval.
FundamentalsRetrieval & Context
Tokenization
The process of breaking input text into smaller units, called tokens, before a language model processes it.
Fundamentals
No concepts in this category yet.