Il glossario dell'intelligenza artificiale e degli agenti
Definizioni chiare e semplici del vocabolario dell'IA e degli agenti in cui continui a imbatterti: agente AI, AI agente, MCP, codifica delle vibrazioni, finestra di contesto e altro ancora. Ogni voce risponde direttamente alla domanda, quindi approfondisce il livello, con collegamenti alle guide e alle lezioni che utilizzano il termine.
AEO (Answer Engine Optimization)
AEO (answer engine optimization) is optimising content so AI answer engines like AI Overviews, ChatGPT and Perplexity return it as the direct answer.
Termine · 2 minimo lettoAgent Harness
An agent harness is the scaffolding around an AI model that runs the loop, manages context, dispatches tool calls and enforces safety so the model can act.
Termine · 3 minimo lettoAgentic AI
Agentic AI is AI that acts autonomously toward a goal, planning and using tools across many steps. Definition, how it differs from generative AI, and real examples.
Termine · 2 minimo lettoAgentic Coding
Agentic coding meaning: writing software by directing an AI coding agent that plans, edits and runs code in a loop, instead of typing every line yourself. Definition, tools and how to start.
Termine · 3 minimo lettoAgentic Engineering
Agentic engineering meaning: building real, production-grade software by directing AI coding agents that plan, edit and run code, while you own the goal, context and verification.
Termine · 3 minimo lettoAI Agent
AI agent meaning, in plain English: a system that uses a language model to decide and act in a loop toward a goal, calling tools along the way. Clear definition, examples, and how AI agents differ from a chatbot.
Termine · 3 minimo lettoAI Guardrails
AI guardrails are the rules and checks that keep an AI system inside safe, allowed behaviour by filtering inputs and outputs and limiting what it can do.
Termine · 3 minimo lettoAI Hallucination
AI hallucination means an AI model stating something false or made-up as if it were true and confident. Meaning, why language models hallucinate, and how to reduce it with grounding and verification.
Termine · 3 minimo lettoAI IDE
An AI IDE is a code editor with a built-in AI coding agent that can read your whole project, write and edit code across files, and run tasks in the editor.
Termine · 2 minimo lettoContext Window
A context window is the maximum amount of text, measured in tokens, an AI model can consider at once, including the prompt, history and the answer it writes.
Termine · 3 minimo lettoEmbedding
An embedding turns text, code or images into a vector of numbers that captures meaning, so a machine can measure how similar two things are.
Termine · 3 minimo lettoFine-Tuning
Fine-tuning means further training a pretrained AI model on your own examples so it learns a specific style, format or task. Meaning, how it differs from RAG and prompting, and when it is worth it.
Termine · 3 minimo lettoGEO (Generative Engine Optimization)
GEO (generative engine optimization) is optimising your content so AI tools like ChatGPT, Gemini and Perplexity cite, mention and recommend it in answers.
Termine · 2 minimo lettollms.txt
llms.txt is a Markdown file at your site root that gives AI systems a clean, curated map of your key pages so they can understand and cite your content.
Termine · 2 minimo lettoMCP (Model Context Protocol)
MCP (Model Context Protocol) is an open standard that connects AI models to external tools and data through a common interface, often called the USB-C for AI.
Termine · 3 minimo lettoMulti-Agent System
A multi-agent system splits a job across several AI agents that each own a role and coordinate, often led by an orchestrator. Meaning, how it differs from a single agent and subagents, and when to use it.
Termine · 3 minimo lettoPrompt Caching
Prompt caching stores the processed prefix of a prompt so repeated requests reuse it, cutting cost and latency. Cache reads can be about 90 percent cheaper.
Termine · 3 minimo lettoPrompt Engineering
Prompt engineering is the craft of writing clear instructions and context so an AI model reliably produces the output you want. Meaning and core techniques.
Termine · 3 minimo lettoRAG (Retrieval-Augmented Generation)
RAG (retrieval-augmented generation) gives an AI model relevant documents to read at answer time, so its response is grounded in your data instead of memory. Meaning, how it works and RAG vs fine-tuning.
Termine · 3 minimo lettoSubagent
A subagent is a specialised AI agent a main agent delegates a task to, running in its own context window with its own prompt and tools, returning a summary.
Termine · 3 minimo lettoSystem Prompt
A system prompt is the standing instruction that sets an AI model role, rules and behaviour before any user message, shaping how it responds all session.
Termine · 3 minimo lettoTool Calling
Tool calling (aka function calling) is when an AI model or agent outputs structured JSON to ask your code to run a function, so it can act and use tools instead of only replying with text. Definition, how it works, and how it relates to tool chaining and MCP.
Termine · 3 minimo lettoVector Database
A vector database stores embeddings and finds the ones closest in meaning to a query, which is how RAG and semantic search retrieve relevant text fast.
Termine · 3 minimo lettoVibe Coding
Vibe coding is building software by describing what you want in plain language and letting an AI write the code, focusing on the result over the syntax.
Termine · 3 minimo lettoWorkflow Automation
Workflow automation uses software to run a multi-step process automatically, triggered by an event, so repetitive tasks happen without manual work each time.
Termine · 2 minimo letto