Glossary
Plain-language definitions of the terms you'll come across in Lorito.
A quick reference to the words used throughout Lorito and this documentation. Where a term has a deeper guide, it links to it.
Core building blocks
Project : Your knowledge base and the home for everything tied to one site or client — its content, its agent, and its channels. Agencies typically run one project per client. See Core concepts.
Knowledge base : The collection of content a project has indexed. It's what your agent draws on to answer questions. Each project has its own knowledge base.
Source : Something you add to the knowledge base — a website, a PDF, or a piece of text or markdown. Each source is read and indexed so your agent can use it. See Adding sources.
Item : An individual piece of content within a source — for example, a single page indexed from a website. A website source can contain many items.
Indexing : The process of reading a source and turning it into searchable content. Lorito does this automatically when you add a source, and it runs in the background.
Embedding : A numeric representation of a piece of content's meaning, created during indexing. Embeddings are how Lorito finds the most relevant content for a question — they let it match by meaning, not just keywords.
Chunk : A small, searchable piece a source is broken into during indexing. Your knowledge capacity is measured in chunks behind the scenes, though you only ever see it as a fill level.
Retrieval (RAG) : How your agent answers using your content. When a visitor asks something, the agent retrieves the most relevant pieces from your knowledge base and answers from them. This is often called RAG (retrieval-augmented generation), and it's what keeps answers grounded in your own content. See Knowledge search.
Your agent
Agent : The assistant that answers your visitors, built from a project's knowledge base and given a persona, instructions, and a model. See Core concepts.
Persona : How your agent presents itself — its name and the voice it speaks in. See Persona & model.
System prompt : The underlying instructions that guide how your agent answers. Lorito drafts these from your content when you set up, and you can edit them at any time. See System prompt.
Model : The AI that generates your agent's replies. You can choose which model powers your agent, or leave it on the default. See Persona & model.
Channels and surfaces
Channel : A surface where your agent answers — where people reach it. Your agent's knowledge and persona stay the same across every channel. See Channels overview.
Deployment : A specific instance of your agent running on a particular channel. The same agent can have several deployments live at once.
Widget : A floating chat launcher you add to your website with a small script snippet. See Website widget.
Iframe : An inline embed that places the chat directly inside a page's layout, rather than floating over it.
Hosted page : A full-page chat hosted by Lorito at your own address, with no website needed — handy as a shareable link.
Custom domain
: Serving the chat from a domain you own, such as chat.example.com, instead of the default Lorito address. Availability depends on your plan.
Tools and skills
Toolkit : A connected external app or service (through an integration) that groups together the actions your agent can take with it. See Tools.
Tool : A single action your agent can take — searching your knowledge base, calling a webhook, or an action from a connected toolkit.
Skill : Reusable instructions you give your agent for handling a particular kind of task. The agent loads a skill when it's relevant to what's being asked. See Skills.
Conversations and billing
Conversation : A single visitor's chat with your agent — the back-and-forth of questions and answers. Lorito records conversations so you can review what people are asking. See Reading conversations.
Credit : The unit Lorito uses to meter your agent's usage. Your agent draws from your plan's credit allowance as it answers. See Credits.
Entitlement : A feature your current plan unlocks — for example, organisation branding, custom domains, or integrations. See Plans & entitlements.