In this article, you will learn how to create a chat configuration in the Backoffice, walk through the sections of the configuration dialog, save it as a draft or publish it directly, and roll it out to your studios.
Contents
- Prerequisites
- Configuration overview
- Creating a new configuration
- Save the configuration
- Detail view and status
- Customer self-service configuration
- Activate and whitelist
- Embed the chat widget
- Third-party data and licensing
Prerequisites
- Permission: Configure MagicAI chat settings, available under Settings / Permissions in the Artificial Intelligence subcategory. A separate permission, MagicAI chat conversations, controls visibility of the agent-side inbox and lives under the Communication subcategory.
- Licensing: MagicAI Chat is not part of the regular pricing tiers; it ships under its own licensing model. Contact your Sport Alliance account manager to enable it for a studio.
- Avatar images: a main chatbot avatar for the open chat widget and a call-to-action avatar for the floating chat button, each up to 3 MB.
- At least one knowledge source: file (max. 10 MB), URL or Zendesk Help Center.
- For a Zendesk support escalation: Zendesk subdomain, agent username and API token.
- For membership offers: a maintained offer channel (Sales tool, Studio pages and Connect API, MySports or Open API).
- Eligible default billing studio: the configuration's default billing studio must be a studio where the chatbot is contractually available and that is part of the configuration's whitelisted studios. If either condition is not met, the configuration cannot be saved.
Configuration overview
Open the configuration entry point at Settings / MagicAI / Chat. The sidebar shows the new MagicAI section with a Chat entry and carries a New badge at launch.
The overview page lists all chat configurations for your location.
Columns in the list:
- Name: the internal identifier, only visible to Backoffice users. A Draft chip appears next to the name while the configuration is still being assembled.
- Brand name: your studio's brand name. Not member-facing on its own, but useful for distinguishing configurations across brands.
- Chatbot name: the public-facing name members see in the chat widget.
- Last updated: the timestamp of the last change to the configuration.
- Published for: a badge with the number of studios the configuration is currently active in. An Active chip appears here once the configuration is activated in at least one studio.
- Facilities: a badge with the number of studios the configuration is whitelisted for; hover or click to see the studio list.
From the list you can filter by status, search by name or brand name, sort and toggle columns. Use the location switcher in the top right to see configurations from other locations. Start the configuration dialog with Create chat configuration in the top right.
Activation and deactivation always run from this overview page. The detail view manages content and whitelisting only. See Activate and whitelist for the full activation rules.
Creating a new configuration
Create chat configuration opens a single dialog that walks you through five sections, each ending with a Next button. The dialog header offers a Save draft action that you can use at any point to persist your work in progress. No step numbers are displayed; the section title and progress within the dialog indicate where you are.
General information
The first section captures the base data of the configuration.
Names
- Name: internal name of the configuration (max. 30 characters). Only visible to Backoffice users.
- Brand name: your brand name (max. 30 characters). Shown in the Backoffice overview.
- Chatbot name: the name members see in the chat widget (max. 30 characters).
Welcome message
- Welcome Message: the greeting shown when a member opens the chat (max. 400 characters).
Billing studio
- Default billing studio: the fallback studio used for chat conversations that cannot yet be assigned to a specific studio (for example, an unauthenticated prospect). The studio must be on the configuration's whitelist and a studio where the chatbot is contractually available; otherwise the configuration cannot be saved.
Legal consent
If you fill in both fields, the preview shows a consent dialog with the actions Decline and Accept that members see before their first message. Both fields are optional: if you leave them empty, no consent dialog is shown.
- Legal consent title: title of the consent dialog (max. 30 characters).
- Content of legal consent: the body text explaining what the member consents to, with rich-text options (bold, italic, underline, text colour, link).
Suggested consent text
The following text was provided by Legal (Bjoern Baltbardis) as a working draft for the consent dialog. Adapt the placeholder links to your own privacy notice and terms of use before going live.
- Title: Data processing notice
- Description: Your input, including personal data, is processed automatically to answer your request. You can withdraw this consent at any time with effect for the future. Please do not enter health data or other sensitive information. You can find more information on data processing and your rights in our [privacy notice] and our [terms of use].
Styling
Customise the visual appearance of the chat widget to match your brand. You configure three base colours plus a bubble shape; all other UI colours (text, borders, error states) are derived automatically.
Colors & look
-
Highlight color: main accent colour as a hex value, used for primary UI elements such as buttons and highlights, e.g.
#166cb8. -
Chat background: chat widget surface colour as a hex value, e.g.
#6d2929. - Header gradient color: optional second colour used for the header gradient. If empty, the header uses the highlight colour. Hex value.
- Use rounded shape for chat bubbles: yes/no toggle.
Avatar
Two avatar slots are supported. Only one image is allowed per slot.
- Main chatbot avatar: the primary logo displayed in the chat widget header (max. 3 MB).
- Call-to-action avatar: the logo or icon shown on the floating chat button that opens the widget (max. 3 MB).
Capabilities
Enable the capabilities the bot should handle. Each capability unlocks specific functionality for end users. After saving you can fine-tune each workflow on its card in the detail view.
| Capability | What it does |
|---|---|
| Studio information | The chatbot can answer questions about studio details (opening hours, address, facilities). |
| Trial session booking | End users can book a trial or test training session through the chatbot. |
| Course discovery | The chatbot can show the classes and courses available at a specific studio within a chosen date range. |
| Contract cancellation | Members can initiate a contract cancellation via the chatbot. |
| Support escalation | When the chatbot cannot resolve a request, it creates a support request and forwards it to a human agent. Requires additional Support escalation configuration (see below). |
| Customer self-service | Members can view and modify their account data (master data, address, contact details, freeze periods, contract data) through the chatbot. Requires additional Self-Service configuration (see Customer self-service configuration). |
| Membership offers | The chatbot can present and recommend membership offers to prospects. Requires selecting a Rate Bundle Group Type. |
Configuring Support escalation
If you enable Support escalation, choose the transfer Type. Only one Support escalation configuration is allowed per chatbot.
Type Email
- Email: recipient address for transferred conversations. Must be a valid email address.
Type Zendesk
- Zendesk URL: URL of your Zendesk instance.
- Zendesk username: the agent username used to create tickets.
- Zendesk token: API token for authentication.
Configuring membership offers
If you enable Membership offers, you must also select the Rate bundle group type the offers are pulled from. Without a selection, the configuration cannot be saved. The four options map to the underlying RateBundleGroupTypeEnum values:
- Sales tool: offers from the Sales tool rate bundle group.
- Studio pages and Connect API: offers from the Studio pages and Connect API rate bundle group.
- MySports: offers from the MySports rate bundle group.
- Open API: offers from the Open API rate bundle group.
Your selection determines where the bot pulls offers from and where the sign-up takes place.
Knowledge sources
At least one knowledge source is required. Use + Add knowledge source to add more. For each source you pick the Source type: File, URL or Zendesk.
Adding a source is only half the job. How you structure the content inside your sources directly affects answer quality, and some topics should be left out entirely because the software already delivers them. For what to include, what to leave out and how to format your content, see Building the knowledge base for MagicAI Chat.
Type File
Upload a text or PDF file directly (max. 10 MB per file). For multiple files, create multiple sources.
Type URL
Enter the starting URL of a web page. The system crawls up to 5 levels deep and follows a maximum of 50 links from the starting URL. Both limits are fixed and cannot be changed.
Type Zendesk
Enter the base URL of your Zendesk Help Center, for example https://yourstudio.zendesk.com. Articles are fetched in all languages available in your Zendesk Help Center, as configured per tenant.
Updating knowledge sources
When you save a new knowledge source, it is automatically published and indexed so the bot can use it.
If the content behind an existing knowledge source changes (for example, your website was updated or you uploaded a new file), trigger an immediate update with the Update knowledge sources button at the bottom of the knowledge sources card on the detail page. This re-indexes all sources right away, so changes appear in the bot's answers immediately. If you do not run a manual update, all knowledge sources are automatically re-indexed every night.
Recommended content and format
Whether the chatbot can answer well depends much more on the structure and language of your sources than on the source type. Use the following recommendations when you prepare content.
- Preferred formats: Markdown or plain text. Both are clearly structured and split into chunks cleanly. PDFs work as long as the text is extractable; avoid scanned PDFs, image-only files and Excel sheets.
- Use an FAQ structure: heading is the question, body is a self-contained answer. Each section must make sense on its own; some repetition between sections is fine if it keeps each block self-sufficient.
- Consistent terminology: use the same words for the same concept throughout. Avoid marketing language, contradictory statements between sources, and outdated content.
Save the configuration
The dialog offers two ways to leave the create flow:
- Save draft (action in the dialog header): persists the configuration in Draft state. Required fields can be empty in this mode. Use it to keep work-in-progress configurations without making them activatable.
- Next (the primary button at the bottom of each section): advances through the sections. On the last section it becomes the submit action and saves the configuration in Published state. Required fields must be complete; otherwise the dialog blocks submit. Published only means the configuration is stored and complete; it does not yet make the chat live for members. Activation in studios is a separate step (see Activate and whitelist).
In both cases you land on the detail page, where each section appears as a card and can be edited individually.
Detail view and status
The detail page has two tabs:
- Settings: cards for General information, Legal consent, Styling (with sub-cards Colors & look and Avatar), Capabilities, Knowledge sources and Code snippet.
- Self-Service: cards for Personal member data and Contract data. See Customer self-service configuration.
The detail page surfaces the current status in two places.
Status chip next to the configuration title:
- Draft: the configuration is still being assembled and cannot be activated.
- Active: published and activated in at least one studio.
- Inactive: published but not currently activated anywhere.
Alert banner at the top of the page:
- Drafts show: "This configuration is not yet complete and is currently in 'Draft' status. To activate it, you must first fill in all required fields."
- Published configurations show a confirmation banner (info severity).
The Edit settings action in the detail-page header is only available when the configuration is in the Published state.
Customer self-service configuration
When you enable Customer self-service in the Capabilities section, the Self-Service tab on the detail page becomes the place to control which account areas members can read and edit through the chat. The tab contains two cards.
Personal member data
- Master data: whether members can view or change their master data.
- Address: access to address information.
- Contact details: access to phone, email and similar contact data.
Contract data
- Contract data: which contracts are visible (All contracts or Active contracts).
- Cancellation of main membership: whether members can cancel their main membership through the chat.
- Suggest freeze period before cancellation: whether the chat proactively suggests a freeze period when a member tries to cancel.
- Freeze periods: whether members can request a freeze period.
Each editable field shows its current status (for example, Activated) and a Check changes link that opens the verification settings for the field.
Activate and whitelist
Two separate concepts control where a configuration is available and where it is activated:
- Whitelisting: the studios that have access to a configuration. Set in the configuration's detail view.
- Activation: the studios where this configuration is the active one. Set from the configuration overview. The chat only becomes visible to members once the embed code is also present on the studio website (see Embed the chat widget).
To activate a configuration:
- Open the overview at Settings / MagicAI / Chat.
- Select the configuration you want to activate.
- Pick the studios where it should run.
Important rules:
- Only one configuration can be active per studio. Activating a configuration in a studio automatically deactivates any configuration that was previously active there.
- If you activate a configuration in a studio that is not yet whitelisted, the studio is added to the whitelist automatically.
- You cannot activate a configuration that is still in draft state. Save it as a complete, published configuration first; otherwise activation fails with Cannot activate a draft configuration.
- You cannot delete a configuration while it is active in any studio. Deactivate it everywhere first.
Embed the chat widget
The detail page contains a Code snippet card with the embed code for your studio website. Copy the snippet and paste it into the source of your website to display the chat widget.
The card behaves differently depending on the configuration status:
- Draft: the card shows the message "The current configuration is currently in draft mode. To obtain a code snippet, please complete the chatbot configuration." The copy action is disabled.
- Published: the snippet is shown together with a Copy action that places the code on the clipboard.
Third-party data and licensing
For studio-location data we use geo data from GeoNames.org, licensed under CC BY 4.0. The licence requires us to disclose the use transparently; please leave this notice in place when you reuse the article.