These eleven concepts show up across every product guide. Understanding them once saves re-reading the same explanation in every manual — each one links to the guide where the full procedure is documented.
Account (tenant)
Section titled “Account (tenant)”Your agency or business inside Nuntius. Every account is an isolated compartment: every piece of data — conversations, leads, documents, keys — is always filtered by your account’s id, and an id from another account on any route returns “not found”, never “not authorized” (which would confirm the data exists). See Security and compliance.
AI agent
Section titled “AI agent”The configurable identity — instructions, model, assigned resources — your assistant uses to talk to customers. You can have several agents handling different things (one for sales, another for support); every channel or channel instance has one assigned. See AI agents.
Channel and instance
Section titled “Channel and instance”A channel is the type of path your agent talks through: WhatsApp, Telegram, Web widget, Messenger / Instagram Direct. An instance is one concrete connection of that type — you can have several instances of the same channel (two widgets, two WhatsApp numbers), each with its own assigned agent, its own responsible people, and its own origin attribution. See Channels.
Conversation
Section titled “Conversation”The message thread with a customer on a channel instance. It has three modes: AI (the agent answers), HUMAN (someone on your team took control), and QUEUED (the AI escalated and is waiting for someone to pick it up). See Conversations.
Lead and contact
Section titled “Lead and contact”A lead is an opportunity in your sales funnel, with a stage, a quality rating and an owner; it’s captured automatically on any channel or entered by hand. A contact is the customer’s own record (phone, verification, marketing lists), shared across all their leads and conversations — it also includes their memory, the summary the agent keeps about them and shares across channels. See Leads and Contacts and marketing.
Resource and tag
Section titled “Resource and tag”A resource is anything an agent can use: a knowledge-base document, a template, or an MCP server. A tag is the mechanism that decides which agent sees which resource — a single catalog, shared across your whole account. Without tags or “Access to everything”, an agent sees no resource at all: denied by default. See AI agents, resources section.
Audience: Internal / Both
Section titled “Audience: Internal / Both”Every action, template or document declares whether it’s Internal (your team only) or Both (also
an anonymous visitor on a public channel, such as your website’s widget). “Internal” is the default for
anything not explicitly declared — fail-closed. A Both-audience piece that writes to an external system
returns an anonymous visitor only the minimal confirmation, never the full record. See
Agent control: resources and permissions.
Permissions and access media
Section titled “Permissions and access media”For an identified caller (a person with a session, an employee with their own channel, an API key), a
fine-grained permission catalog decides what they can do — always intersected with their access
medium’s ceiling (an API key’s sealed scopes, or a channel’s user_medium_grants cut-down). See
Agent control: resources and permissions and
API keys and access media.
Job, flow and instruction
Section titled “Job, flow and instruction”A job is a standalone task your agent runs, once or repeated. A flow chains several steps with the option to branch on a condition. An instruction (“consigna”) is a flow you ask your assistant to build over chat, in your own words — it can be edited exactly like one built by hand. See Jobs and flows.
API key
Section titled “API key”A secret credential for calling /public/v1 from your own code — built for machine-to-machine
integrations, never for a person to log in with it. It’s shown in full only once; its read/manage scopes
per resource are sealed the moment it’s created. See
API keys and access media and Authentication.
Model Context Protocol, the open standard for an AI assistant to use external tools in a structured way. It runs in two opposite directions in Nuntius: Nuntius’s own MCP server exposes your account (leads, contacts, appointments…) to YOUR external AI assistant (Claude Desktop, Claude Code, Cursor); an external MCP server (BYO-MCP), which you register in Settings, plugs third-party tools into the agent that talks to your own customers — under your own responsibility. See Integrations and MCP.
See also
Section titled “See also”- Glossary — every term, in one line, alphabetically.
- Security and compliance — the four access layers, explained in full.