How an assistant works

There’s no single “AI chat box” in Intric. Instead, your organization can create as many assistants as it needs, each one built to be good at a different thing — a procurement assistant, a legal expert, an HR support assistant. Every assistant is the same three ingredients combined differently.

A model, instructions, and knowledge

The language model

The model is what actually processes language: it reads the instructions and the knowledge you’ve given the assistant, and generates the response. Some models are fast and general-purpose; others reason more carefully through complex questions. Your organization decides which models are available, and you pick the one that fits the task.

The instructions

Instructions tell the assistant how to behave: its role, its tone, and what it should and shouldn’t do. A general-purpose assistant might be told to answer plainly and stay within a certain topic. A specialized one might be told to act as a strict proofreader or a meticulous case handler.

The knowledge

Knowledge is what the assistant can draw on beyond its own training — your organization’s documents, connected systems, or the live web. An assistant with no attached knowledge answers from general understanding alone; one with the right knowledge attached answers from your organization’s own material.

Combine the right model, clear instructions, and relevant knowledge, and you get an assistant that gives precise, grounded answers rather than a generic one.

The personal assistant

Every user gets a personal assistant, and it’s what you see the moment you sign in. It’s meant for everyday work that doesn’t need to be shared: a quick question, a draft email, a summary of something you were just sent. You don’t configure it yourself — your organization does, centrally, for everyone at once. See Personal assistant.

Shared assistants in Spaces

Beyond the personal assistant, your organization can create any number of specialized assistants inside Spaces — shared work areas for a team or a department. A shared assistant is built once and used by everyone with access to that Space, so one person’s work configuring it becomes a resource for the whole team.

Why the same question gets different answers

Ask the same question to your personal assistant and to a specialized assistant in a Space, and you can get very different answers — by design. The personal assistant answers from general knowledge and whatever web search or connected tools your organization has enabled for it. A specialized assistant answers from the specific knowledge and instructions it was built with. Neither is wrong; they’re built for different jobs.