Schoolbox AI Services

AI tools in Schoolbox help staff create content faster, support student learning and reduce time spent preparing materials. 

Instead of switching between external AI tools and Schoolbox, staff can use AI directly inside the platform while creating content, planning lessons and building learning activities.

Common uses include:

• improving written content
• generating lesson plans
• creating learning resources
• facilitating structured student discussions
• generating images for lessons and communications

All AI features are optional and can be enabled or configured by your school’s administrator.

Schoolbox currently includes several AI-powered tools designed to support teaching and content creation.

Additional AI-powered features may be added as Schoolbox evolves.

Use the navigation panel on the right to learn how to enable or use each feature.

Schoolbox integrates with external AI providers to generate responses and content. Schools can connect their own AI service and control which model is used.

Bring Your Own AI
Schoolbox uses a Bring Your Own AI (BYO AI) approach. This means your school connects its own AI provider, such as:

• OpenAI
• Microsoft Azure OpenAI
   

This allows your school to manage:

• Usage costs
• Data policies
• AI model selection
• Response behaviour through reasoning effort and text verbosity settings

Note: Some AI models only support basic editing features. If a model does not support structured outputs, advanced features such as Lesson Plan generation may not be available. Reasoning effort and text verbosity settings are only available for supported models and providers.

Advanced settings

Supported models can also be configured using the following options:

• Reasoning effort controls how much reasoning the model performs before generating a response.
• Text verbosity controls how concise or detailed generated responses should be.

Both settings can be configured independently for each Schoolbox AI feature. If left as default, the provider's default behaviour will be used.
 

Data and Privacy

AI responses are generated using a third-party AI provider configured by your school. Schools should review their provider’s privacy and data policies before enabling AI features.

Superusers have the ability to configure AI services to be used throughout their instance on a per feature basis.

1. Go to Administration > System Settings > AI Services

2. Identify the feature to configure.

3. Select the provider

• OpenAI
• Azure

Note: If 'Off' is selected the feature will be turned off for your instance.

4. Add the API key, resource name and deployment ID from your third-party AI provider. [See setup provider instructions]

5. [If Applicable] Set a Generation System Prompt that allows for school level guardrails and guidance to be added to all AI interactions with that feature.

6. [If Applicable] Enable Inline AI Moderation - this is recommended, however note that content is processed by a third-party AI service outside school control.

Superusers have the ability to review logs of AI interactions that have occurred within Schoolbox.

1. Go to Administration > System Settings > AI Usage Log

2. Filter by keyword, feature, user and date.

3. Click the    to view the interaction log.

Note: Logs can currently only be viewed for image generation and discussion activity within Schoolbox. Further logs can be viewed from the administration portal of the configured LLM. 

Pseudonymisation

Schoolbox uses pseudonymisation to reduce the identifying information sent to an external AI provider, such as OpenAI or Azure, when generating Differentiation Profiles.

The process replaces detected names and selected identifiers with substitute values called “tokens”. Information about the student’s learning and support needs remains available to help generate a useful profile.

Pseudonymised information is still private information. Names that are not detected, and other details within a record, may still identify someone.

How Schoolbox identifies personal information

Schoolbox combines known information from school records with automated detection.

Known-person matching uses available details for:

• The student: full, first, preferred and last names, student identifiers and date of birth.
• The student’s parents: names available through SIS contacts and linked Schoolbox parent accounts.
• People directly linked to the submitted pastoral records: authors, people who last modified the records, individually listed users with visibility, and other students involved through linked records.

Matching covers different written forms, including capitalisation, reversed name order, underscores and joined names. For example, it can match Alex Morgan, Alex_Morgan, AlexMorgan and Morgan, Alex.

Automated detection uses Presidio (originally developed by Microsoft) to help identify other names, locations, phone numbers, email addresses and Australian Medicare numbers in the prepared text.

What content is included?

• Confidential content is excluded. The Confidential tab and confidential attachments of a Pastoral Care record are not included in this process.
• General content may be included. General text and supported general attachments from records contributing to the profile can be used.
• Supported attachments include PDF, DOCX, DOC and ODT files. Unsupported files and attachments with no extractable text, such as image-only scans, are skipped. If a supported attachment cannot be read, generation stops and staff are asked to resolve the problem.

What are the limitations?

Schoolbox does not guarantee that every name or identifier will be removed.

Known-person matching depends on the details available in Schoolbox and the SIS. Other people mentioned in the text, including external practitioners, may rely on automated detection. Spelling differences, unfamiliar names and identifying details in the surrounding text can affect coverage.

Matching may also replace ordinary words that are someone’s name, such as “May” or “Young”. This can reduce the clarity of some content.

The AI provider still receives sensitive information about the student’s learning and support needs. Staff should include only relevant information, avoid unnecessary personal details and review the generated profile before using it.

What do schools need to do?

Schools need to configure their connected AI service to meet their privacy requirements. This includes ensuring submitted information is not used to train AI models, and reviewing:

• How long information is retained.
• Who can access it.
• Where it is processed.

These controls depend on the provider, service and account arrangements. Schoolbox’s pseudonymisation process does not configure or replace them.

Does Schoolbox have an AI Charter?
Yes, the AI charter can be found here.

I have received the following error "xxx does not have to the model 'xxx'". What do I do?
This means that the API key provided does not have access to the model selected in the drop down in Schoolbox.  In OpenAI the allowed models are configured on the Project Limits screen.