Executive Summary

Brillium recently experienced a temporary service disruption affecting the AI Essay Grading capability within Assessment Builder. This disruption occurred during a required upgrade to our underlying AI model after the previous model was deprecated by our cloud provider.

The issue has been fully resolved, services have been restored, and there was no impact to customer data. We take full responsibility for the disruption and have implemented additional safeguards to strengthen our change management and service reliability processes.

‌

Impact Overview

• Service Affected: AI Essay Grading (Assessment Builder)
• Customer Impact: Temporary unavailability of automated grading
• Workaround Provided: Temporary routing to manual grading
• Data Integrity: No loss, corruption, or unauthorized access to customer data

All candidate submissions and assessment results remained secure, complete, and unchanged throughout the incident.

‌

Root Cause

The disruption was the result of an external dependency change. The Large Language Model (LLM) previously used by Brillium for AI Essay Grading was deprecated by AWS, requiring an accelerated transition to a supported next-generation model.

During this transition:

• Automated grading was temporarily paused to ensure a controlled and secure upgrade
• Additional validation was required to confirm grading accuracy, consistency, and fairness
• Safeguards were applied to protect processing integrity and prevent inaccurate scoring

While this action was necessary to maintain platform stability and vendor support compliance, it introduced a temporary service interruption while validation and stabilization were completed.

‌

Resolution

Our engineering team has successfully:

• Transitioned the AI Essay Grading engine to a supported next-generation LLM
• Completed comprehensive regression testing and quality assurance
• Validated grading precision, consistency, and fairness across all scenarios
• Fully restored automated scoring for all assessments

Manual grading processes introduced during the incident have now been discontinued.

‌

Current Status

• AI Essay Grading is fully operational
• All assessments are processing normally
• No further action is required

‌

What We Are Doing to Prevent Recurrence

We are strengthening our platform and processes to reduce the likelihood and impact of similar events:

• Enhanced Third-Party Dependency Monitoring
  Improved tracking of vendor lifecycle changes, including deprecations and end-of-support timelines
• Proactive Upgrade Planning
  Earlier adoption cycles to reduce the need for accelerated transitions
• Expanded Pre-Deployment Testing
  Increased simulation of production scenarios prior to release

‌

Customer Assurance

We recognize how critical reliable scoring is to your operations. This upgrade ensures:

• Improved grading accuracy and consistency
• Greater long-term system stability and vendor alignment
• Continued protection of all customer data

Brillium remains committed to delivering secure, high-performance, and future-ready assessment solutions.

‌

Support

If you have any questions or need assistance, our support team is here to help.