The Steeple AI Assistant — overview and usage

Overview

TheAI Assistant is a tool integrated into Steeple that lets you interact with the platform through conversation. It can answer questions about how Steeple works and perform actions on your behalf (create an event, search for a coworker, create a post, etc.).

The feature is in beta.

Access the assistant

Two entry points, which share the same conversations:

• the side panel, opened from the Steeple interface; it remains available while navigating the platform;

• theAssistant app full page, which provides access to conversation history and recipes.

Choose a scope

The assistant works by scopes (scopes). Each scope provides access to specific features:

• Administration : manage users, departments, locations, communities, TV screens

• Directory : search for coworkers, view distribution by department or location

• Posts : manage posts and sections

• Calendar : view events, create them, track participation

• Shortcuts : manage shortcuts to links, documents, or media

• Feeds : manage automatic or manual content feeds

• Jobs : manage offers and track applications

• Notifications : view notifications and configure preferences

• Messaging : view conversations and send messages

• Documents : answer from document content, browse and search

• Profile : view and edit your account information

The details of the possible actions for each scope are described in the article Scopes and available actions.

Send a message

• Type the message in the text field at the bottom of the panel

• Press Enter to send (or Shift+Enter for a line break)

• The assistant displays "Thinking..." while it processes the request

Attach a document

It is possible to attach a document or an image so the assistant can use its content:

• Click the paperclip icon next to the input field

• Select a file (PDF, Word, Excel, PowerPoint, OpenDocument, CSV, TXT, RTF, HTML, Markdown, EPUB, or image)

• The content is analyzed and sent to the assistant with the following message

The details of accepted formats, maximum sizes, and use cases are described in the article Attach a document to the assistant.

Action confirmations

When the assistant is about to perform a write action (create, modify, or delete something), a confirmation card is displayed with the details of the planned action. The fields are editable in place : title, date, audience, content — you can edit directly in the card before validating, without having to rephrase your request.

Two buttons then allow you to accept or refuse. The assistant never performs a modification action without your approval.

Actions performed

After each action, a collapsible summary indicates what was done (for example "1 action performed"). Click it to see the details of the operations carried out.

Navigation

The assistant can suggest direct links to other Steeple pages in the form of "Go to [Page]" buttons. Click them to navigate directly.

Suggestions

When a conversation opens, the assistant displays a welcome message with suggested questions tailored to the active scope. Click a suggestion to send it directly.

Give feedback

On the assistant's latest message, buttons thumbs up and thumbs down let you rate the response. An optional comment can be added to provide more detail.

New conversation

To change topic, click on New conversation when the assistant is not processing a request. The previous conversation is not lost: it is added to the history.

Conversation history

Conversations are kept and listed on the My conversations page of the Assistant app. Each one automatically gets a title based on its content.

• Click on a conversation to resume it where it left off

• Older exchanges load as you scroll

• A conversation can be deleted individually

The current conversation is also restored after a page reload or navigation within Steeple.

Response sources

When the response is based on the organization's documents, the assistant displays below its message the sources used : document name and page. Click a source to open the relevant document. See Scopes and available actions for the Documents scope.

Generate an image

The assistant can generate illustrations — for example an image to accompany a post or the cover of an event. It suggests this on its own in these contexts, and you can also ask it to.

• The generated image is displayed below its response

• Requesting an edit by describing it in words (« brighter », « without text ») produces a new version

• The current image remains the active image of the conversation: this is the one the assistant reuses when you refer to « the image » without further detail

• An image can also be imported from the device to serve as a base

Restrict to a community

The assistant can work within the scope of a community specific. When that is the case, a Community: name appears above the input field and the assistant limits its responses and actions to that community. The cross on the chip lets you leave this scope.

Usage limits

The assistant's daily usage is capped per user, and the cap depends on the organization's plan — the paid AI plan provides a significantly larger quota than the basic plan. When the limit is reached, the assistant replies that you need to try again the next day. The counter resets each day.