AI エージェントとは何か、そして何でできているのか
曖昧な言い回しを抜きにした用語の整理です。エージェントとは実際に何なのか、どんなエージェントならお客様の前に出せるのか、コンソールにある各部品が何をしているのか。
まずはここから
それが実際に何なのか。そして「うちはもう GPT を使っている」が、同じ話ではない理由。
What is an AI agent?
Software that can hold a conversation, look things up, and take actions on your systems — deciding for itself which of those to do next.
読むAI agent vs LLM: what is the difference?
The model is the engine that turns text into text; the agent is everything built around it — retrieval, tools, memory, boundaries and channels. Models are replaceable parts inside an agent.
読むWhat is a business AI agent?
An agent that works for a specific business — answering from that company's own documents, inside a scope its owner defines, and accountable for what it says.
読む何でできているか
コンソールで設定する 5 つの要素と、それぞれが受け持つ役割。
What is an agent's soul (system prompt)?
The standing description of who an agent is — its character, tone and non-negotiables — held separately from the job it happens to be doing.
読むWhat is a knowledge base for an AI agent?
The documents an agent may answer from. agent4.io keeps two kinds — your company's vetted material, and each customer's own files — retrieved together but never confused for one another.
読むWhat is an agent Skill?
A named capability pack — a short description of when to use it, fuller instructions, and the tools it unlocks — that the agent loads only when it becomes relevant.
読むWhat is MCP (Model Context Protocol)?
The Model Context Protocol — an open standard for connecting an agent to external tools and data, so a system exposes its capabilities once instead of once per AI product.
読むKnowledge base, tools or Skill — which does what
Three different jobs that get confused for one — the knowledge base holds facts, an MCP tool performs a calculation you cannot let a model improvise, and a Skill is the procedure that decides which to reach for.
読むFine-tuning vs RAG: when you actually need to fine-tune
Continuing a model's training on examples of the behaviour you want, so that behaviour becomes its default instead of something you ask for on every request. It changes how a model works, not what it knows.
読むどう振る舞うか
誰も入力していないとき、そして全員が帰った後に、エージェントが自分で何をしているか。
What is a page playbook (page-aware AI chat)?
A per-page briefing for the agent — background it should already know, an opening line, and a few suggested questions — chosen automatically from the URL the visitor opened the chat on.
読むWhat are scheduled tasks for an AI agent?
Work an agent commits to for a specific future moment — a follow-up it decides to send, or a recurring reminder — executed on time by a scheduler rather than waiting for someone to speak first.
読むWhat are ask forms (forms inside the chat)?
A small form the agent composes mid-conversation — choices, a number, a date — rendered as something to tap rather than a paragraph of questions to answer by typing.
読むWhat is a space (agent workspace)?
The isolated container holding one customer's conversations, documents and memory — enforced at the database level, so one customer's material cannot surface in another's conversation.
読む動かすまで
多くの試験導入が止まるのはここです。そして、止まらずに済む理由。