/
Agent User Guide

Agent User Guide

Introduction

Agent is a browser‑based AI assistant that lets you work with your organization’s data, documents, and external systems through a chat interface. It combines large language models (LLMs) with your internal data and documents so you can ask questions, analyze data, generate content, and automate tasks in a single, secure environment.

This page gives a high‑level picture of how Agent is organized and links to more detailed topics.

Core Concepts

Agent revolves around three core ideas:

  • Chat – Your main workspace for interacting with the AI.

  • Context – The information Agent can use: uploaded files (“Resources”) and connected systems (“Sources”).

  • Automation & Specialization – Reusable setups for recurring tasks (Instructions, Workflows, Chatbots, Multi‑Chatbots).

Chat and Actions

The Chat view is where you interact with Agent via text or voice, choose models, and apply Instructions that define tone, role, and level of detail. From here you can also invoke specialized actions such as /analyze, /translate, /reason, and /process-file.

Context

Agent can use two kinds of context:

  • Resources – Files you upload (and optionally group) for use in chats.

  • Sources – External systems configured by you or your administrators (e.g. Index, Web, SharePoint).

For further details, see:

Automation and Reuse

Agent supports several ways to capture and reuse work:

  • Instructions – Reusable behavioral profiles for the assistant (role, tone, depth).

  • Workflows – Multi‑step Task chains for running structured processes on files and sources.

  • Chatbots – Standalone assistants with their own configuration and knowledge base.

  • Multi‑Chatbots – Bots that route questions to multiple specialized Chatbots based on routing rules.

Accounts, Access, and Configuration

Access to Agent and its features depends on how your organization’s Agent is deployed (SaaS, Azure Marketplace, or Index‑integrated). Administrators control user accounts, roles, available Sources, and retention policies.

For further details, see:

Using This Documentation