The language behinduseful intelligence.

A plain-language atlas of concepts, techniques, infrastructure, models, and applications.

Open any term for the one-sentence meaning, a deeper explanation, practical uses, and the related language that helps the idea click.

AZIndex the language.
Interrogate the source.

Find your next idea

Ask the glossary a real question.

Describe what you are trying to understand. We will surface the three concepts that give you the clearest next step.

Your question

Try asking

Three ideas are waiting

Start with one of the questions above, or use your own words.

Five learning chapters

Find the territory before you choose the route.

Start with the chapter closest to the question in front of you, then follow related terms until the whole system makes sense.

Chapter 01

Fundamentals

The base language for understanding how intelligent systems learn, reason, and respond.

12records

01.01Large Language Model (LLM)A neural network trained on massive text datasets that can generate, understand, and reason about human language.01.02TransformerThe neural network architecture behind all modern LLMs, using self-attention mechanisms to process sequences in parallel.01.03TokenizationThe process of breaking text into smaller units (tokens) that an AI model can process, typically subwords or word pieces.01.04Context WindowThe maximum amount of text (measured in tokens) that an AI model can process in a single request, including both input and output.01.05HallucinationWhen an AI model generates information that sounds plausible but is factually incorrect, fabricated, or not grounded in its training data.01.06Multimodal AIAI models that can process and generate multiple types of data: text, images, audio, video, and code.01.07TemperatureA parameter that controls the randomness of AI model outputs, with lower values producing more deterministic responses and higher values producing more creative ones.01.08Attention MechanismA neural network component that allows models to dynamically focus on the most relevant parts of the input when generating each token of output.01.09Long ContextThe ability of AI models to process and reason over very large inputs: hundreds of thousands or millions of tokens: in a single request.01.10Model CollapseA degradation phenomenon where AI models trained on AI-generated data progressively lose quality, diversity, and accuracy over successive generations.01.11Neural NetworkA computational system inspired by the brain, composed of layers of interconnected nodes (neurons) that learn patterns from data through training.01.12Responsible AIA framework for developing and deploying AI systems that are fair, transparent, safe, privacy-preserving, and accountable.

Chapter 02

Techniques

Methods used to guide behavior, improve output, retrieve context, and evaluate results.

23records

02.01RAG (Retrieval-Augmented Generation)A technique that enhances AI responses by retrieving relevant information from a knowledge base before generating an answer.02.02EmbeddingsNumerical vector representations of text that capture semantic meaning, enabling similarity search and clustering.02.03Fine-TuningThe process of further training a pre-trained AI model on your specific data to improve performance on domain-specific tasks.02.04Prompt EngineeringThe practice of crafting effective instructions for AI models to produce desired outputs consistently.02.05Function Calling (Tool Use)An AI capability where the model can decide to invoke external functions or APIs based on the conversation context.02.06AI AgentAn AI system that can autonomously plan, reason, use tools, and take actions to accomplish goals with minimal human intervention.02.07Chain of Thought (CoT)A prompting technique that improves AI reasoning by instructing the model to break down complex problems into intermediate steps before giving a final answer.02.08Few-Shot LearningA prompting technique where you provide a small number of input-output examples in the prompt to teach the model the desired behavior.02.09Zero-Shot LearningThe ability of an AI model to perform a task based solely on instructions, without any training examples provided in the prompt.02.10Transfer LearningA machine learning technique where a model trained on one task is adapted to perform a different but related task, reducing the data and compute needed.02.11Reinforcement Learning from Human Feedback (RLHF)A training technique that aligns AI model behavior with human preferences by using human feedback to reward desired outputs and penalize undesired ones.02.12Semantic SearchA search approach that finds results based on meaning rather than exact keyword matches, using embeddings to understand the intent behind queries.02.13Structured Output / JSON ModeA feature that forces AI models to return responses in a specific format like JSON, ensuring parseable and type-safe outputs for programmatic use.02.14StreamingA method of receiving AI model output token-by-token in real time as it is generated, rather than waiting for the complete response.02.15Batch ProcessingProcessing multiple AI requests together as a group, typically at lower cost and higher throughput than real-time individual requests.02.16Agentic WorkflowA multi-step AI process where an LLM autonomously plans, executes, and iterates on tasks using tools and feedback loops.02.17AI Hallucination DetectionTechniques and systems for identifying when an AI model generates false, fabricated, or unsupported information that appears plausible.02.18Data LabelingThe process of annotating raw data (text, images, audio) with labels or tags so it can be used to train and evaluate machine learning models.02.19Grounding (AI)Connecting AI model outputs to verifiable sources of truth: such as retrieved documents, databases, or real-time data: to reduce hallucination and increase factual accuracy.02.20Multimodal RAGAn extension of RAG that retrieves and reasons over multiple data types: text, images, tables, charts, and audio: not just text documents.02.21RAG Pipeline (Detailed)The complete end-to-end system for Retrieval-Augmented Generation, including document ingestion, chunking, embedding, indexing, retrieval, reranking, and generation.02.22Synthetic DataArtificially generated data that mimics real-world data, used for training AI models when real data is scarce, expensive, private, or biased.02.23Tool Use (AI)The capability of AI models to interact with external tools, APIs, and systems by generating structured function calls based on natural language instructions.

Chapter 03

Infrastructure

The systems, interfaces, and operational layers that turn a capability into a product.

17records

03.01MCP (Model Context Protocol)An open standard by Anthropic that provides a universal way for AI models to connect to external data sources and tools.03.02Vector DatabaseA specialized database optimized for storing and searching high-dimensional vector embeddings, enabling semantic similarity search.03.03InferenceThe process of running a trained AI model to generate predictions or outputs from new inputs, as opposed to training the model.03.04API GatewayA server that acts as a single entry point for AI API requests, handling routing, rate limiting, authentication, and load balancing across multiple AI providers.03.05Model ServingThe infrastructure and process of hosting a trained AI model and exposing it as an API endpoint for real-time or batch inference.03.06Edge AI / On-Device AIRunning AI models directly on user devices (phones, laptops, IoT) rather than sending data to cloud servers for processing.03.07GPU / TPUSpecialized processors designed for the parallel mathematical operations that AI models require for training and inference.03.08QuantizationA technique that reduces AI model size and memory requirements by using lower-precision numbers to represent model weights, trading a small accuracy loss for major efficiency gains.03.09DistillationA technique where a smaller "student" model is trained to replicate the behavior of a larger "teacher" model, achieving comparable quality at lower cost.03.10LatencyThe time delay between sending a request to an AI model and receiving the response, critical for real-time user-facing applications.03.11Token Limits / Rate LimitingRestrictions imposed by AI API providers on the number of tokens processed or requests made within a given time period.03.12AI GuardrailsSafety mechanisms that constrain AI system behavior, preventing harmful outputs, prompt injection, data leaks, and off-topic responses.03.13AI OrchestrationThe coordination of multiple AI models, tools, and data sources in a unified pipeline to accomplish complex tasks that no single model can handle alone.03.14AI PipelineA sequence of data processing and AI inference steps that transforms raw input into a useful output, typically involving preprocessing, model inference, and post-processing.03.15Inference OptimizationTechniques to make AI model predictions faster, cheaper, and more efficient in production, including quantization, batching, caching, and model distillation.03.16Knowledge GraphA structured representation of information as a network of entities and their relationships, used to give AI systems organized, queryable world knowledge.03.17Token Economy / AI PricingThe cost structure of AI APIs based on token consumption, where pricing is determined by the number of input and output tokens processed per request.

Chapter 04

Models

Model families, how they differ, and the tradeoffs that shape where they fit.

11records

04.01GPTOpenAI's family of generative pre-trained transformer models for language, reasoning, coding, vision, and tool-using applications.04.02ClaudeAnthropic's family of AI models known for long context windows, strong reasoning, and instruction-following capabilities.04.03GeminiGoogle's multimodal AI model family optimized for text, image, audio, and video understanding and generation.04.04LlamaMeta's open-weight model family, available for download and deployment across cloud, data-center, edge, and local environments under its license.04.05Open-Source AIAI models whose weights and architecture are publicly available, allowing anyone to inspect, modify, run, and build upon them.04.06Diffusion ModelA generative AI model that creates images, video, or audio by gradually removing noise from random static, guided by a text or image prompt.04.07Embedding ModelA specialized AI model that converts text, images, or other data into numerical vectors (embeddings) that capture semantic meaning for search and comparison.04.08Foundation ModelA large, general-purpose AI model trained on broad data that serves as a base for many downstream tasks through fine-tuning, prompting, or adaptation.04.09Mixture of Experts (MoE)A model architecture where multiple specialized sub-networks ("experts") are combined, with a gating mechanism that routes each input to the most relevant experts.04.10Transformer Architecture (Detailed)The complete technical architecture of the Transformer, including multi-head self-attention, positional encoding, feed-forward layers, and the encoder-decoder structure.04.11Vision Language Model (VLM)An AI model that can process and reason about both images and text simultaneously, enabling visual question answering, image description, and multimodal analysis.

Chapter 05

Applications

The product patterns and practical jobs these systems can support in the real world.

12records

05.01Computer VisionThe field of AI that enables machines to interpret and understand visual information from images and video.05.02Natural Language Processing (NLP)The branch of AI focused on enabling computers to understand, interpret, and generate human language in useful ways.05.03Text-to-Speech (TTS)AI technology that converts written text into natural-sounding spoken audio, enabling voice interfaces and audio content generation.05.04Speech-to-Text (STT)AI technology that converts spoken audio into written text, enabling voice input, transcription, and voice-controlled interfaces.05.05Image GenerationAI models that create new images from text descriptions (prompts), enabling automated visual content creation.05.06Video GenerationAI models that create video content from text prompts or images, an emerging capability for automated video production.05.07Sentiment AnalysisAn NLP technique that determines the emotional tone of text, classifying it as positive, negative, neutral, or more granular emotions.05.08Named Entity Recognition (NER)An NLP technique that identifies and classifies named entities in text, such as people, organizations, locations, dates, and monetary values.05.09Autonomous AgentsAI systems that can independently plan, execute multi-step tasks, use tools, and adapt their approach based on results, with minimal human oversight.05.10ChatbotAn AI application that conducts conversations with users through text or voice, handling questions, tasks, and interactions in natural language.05.11CopilotAn AI assistant integrated into a user's workflow that provides real-time suggestions, completions, and assistance alongside the user's work.05.12Conversational AIAI systems designed for natural, multi-turn dialogue with humans, maintaining context across exchanges and handling follow-up questions naturally.

From vocabulary to a real decision

A useful term should sharpen the next move.

Bring the outcome, constraint, and risk. We can turn the language into a source-backed evaluation and the smallest credible proof.

Find my opportunity Start a project Compare options with evidence