iBabs Debrief Technical Documentation
Introduction
This technical documentation provides a comprehensive overview of iBabs Debrief, including its system architecture, data flow, security framework, and compliance controls. It describes how the module operates within the Microsoft Azure environment and details the processes that enable secure meeting transcription, summarisation, and document generation.
The purpose of this documentation is to ensure clarity, transparency, and auditability for all relevant stakeholders, including customers, auditors, partners, and supervisory authorities. It demonstrates how iBabs Debrief has been designed and implemented in accordance with iBabs B.V.’s ISO 27001–certified information security framework, while aligning with the broader privacy, security, and AI governance standards of the Euronext Corporate Solutions group.
As part of the Euronext Corporate Solutions (ECS) group, iBabs benefits from shared governance structures, legal frameworks, and group-level compliance oversight, while maintaining independent operational responsibility for its ISO 27001–certified platform, infrastructure, and security management. This ensures that both iBabs’ internal controls and the group’s overarching data protection and AI principles work together to safeguard customer data and ensure compliance with European and United Kingdom regulatory requirements, infrastructure, and security management.
The document outlines how data protection is embedded throughout the system lifecycle — from the moment data is uploaded, processed, and stored to its review, export, and deletion. It highlights the safeguards and technical measures that ensure confidentiality, integrity, and availability of data, while maintaining full customer control and separation of data between environments.
Through this combination of certified operational security and group-level governance, iBabs and Euronext Corporate Solutions together uphold their commitment to providing innovative, reliable, and compliant digital services. iBabs Debrief exemplifies this approach by enabling efficient and secure meeting documentation, powered by transparent, ethically designed, and human-supervised AI technology.
Definitions
The following terms are used throughout this document to ensure clarity and consistency in the description of iBabs Debrief’s technical and compliance framework.
| Term | Definition |
| AI (Artificial Intelligence) | Technology enabling systems to perform tasks that typically require human intelligence, such as speech recognition and natural language processing. |
| AI Model | A pre-trained algorithm provided by Microsoft Azure used to convert speech to text and generate summaries within iBabs Debrief. This Model is more described here: The Whisper model from OpenAI - Azure AI services | Microsoft Learn |
| Azure | Microsoft’s cloud computing platform providing secure data storage, processing, and AI services for iBabs Debrief. |
| Controller | The customer (e.g., meeting organiser or institution) determining the purposes and means of processing personal data. |
| Data Subject | Any individual whose personal data is included in meeting recordings or transcripts. |
| DPIA (Data Protection Impact Assessment) | A process used to identify and minimise data protection risks related to processing activities involving personal data or AI systems. |
| Encryption in Transit / at Rest | Security measures ensuring data is encrypted during transmission (TLS) and while stored (Azure-managed or customer-managed keys). |
| Processor | iBabs, acting on behalf of the Controller, processes personal data solely for service delivery and under contractual instructions. |
| RBAC (Role-Based Access Control) | Access control method used in Azure to ensure that only authorised users and service accounts can access data or resources. |
| Transcript | Textual representation of the meeting audio, generated by the AI model and reviewed by the user. |
Scope of the system
Module Name: iBabs Debrief
Version: 1.0 (initial version – no previous version)
Deployer: iBabs, a subsidiary of Euronext Corporate Solutions, registered in the Netherlands
Purpose: iBabs Debrief is an AI-powered transcript & minutes generating tool, closely integrated within the iBabs environment. The aim is to facilitate and accelerate the admin work for iBabs users, responsible for producing meeting minutes, action and decision lists during and after the meetings.
Short technical description: iBabs Debrief allows assistants and secretaries in charge of providing meeting outputs, record meetings and use the recordings as basis for transcribing and processing further in a form of meeting summary or minutes. The process involves a user's check with each step that involves AI processing or generation of outputs, ensuring the accuracy and human oversight. Once the recording is processed into transcription, the user then can review transcription, edit it, assign the right attendees to the identified speakers, and generate documents for summary and meeting minutes. All three forms of meeting output: full transcript, meeting minutes or summary, can be exported back to connected agenda in iBabs or downloaded in PDF format after user reviewed and edited it. Processed recordings and transcripts are retained only for as long as necessary to provide the agreed services or as instructed by the user. Once deleted, data is securely removed from active and backup storage in accordance with iBabs retention policy.
Forms of availability:
- SaaS module integrated into the iBabs platform, accessible via web browser
- Access through SSO authentication
No standalone distribution or on-premise installation is offered
Implementation Details
Deployment Model: Cloud
System Architecture:
iBabs connector
A collection of private API's to authenticate, authorize, get and store information in iBabs.
The iBabs connector interacts with the following components:
Deployer Core Platform: via private APIs for authentication and document storage
Client systems: PDF export via iBabs APIs
The iBabs connector does not communicate with any other external or third-party AI systems .
File Upload
The user uploads a meeting audio recording to a secure Azure storage location using a provided link (Shared Access Signature, SAS).
Access to the upload and processing environment is restricted through Azure Active Directory–based authentication and role-based access control (RBAC). This ensures that only authorised users and service accounts can access data or system resources. iBabs itself does not access customer data, except with the prior consent of the customer and solely for maintenance or troubleshooting purposes. As a principle, the customer is the only party with access to its own data. All customer data is stored in logically separated environments, ensuring complete segregation between customer tenants and preventing cross-access between datasets.
Data Encryption
Data encryption is implemented across all stages of the system to ensure confidentiality and integrity of user data.
All data is encrypted both in transit and at rest.
To enhance key management, the system adopts an envelope encryption approach, where unique data keys are generated and encrypted with a master key. This ensures secure key rotation, minimizes key exposure, and isolates encryption contexts across data objects.
Automatic Processing
The uploaded file is scanned for security.
The system automatically converts the audio and video into the correct format of audio and generates a transcript.
Speaker separation (diarization) is applied to identify different speakers. The diarization output is stored with minimal metadata and does not include biometric or voiceprint identification data.
Transcript Availability
Once ready, the transcript is stored in a secure location and made available to the user for review and editing.
Errors during processing are automatically handled, and the user will be notified if an issue occurs.
Retention and deletion of transcripts follow the user’s actions within iBabs Debrief and Euronext Corporate Solutions secure data retention procedures.
Editing & Version Control
Users can make edits to the transcript.
Changes can be undone until the user clicks Save.
Once saved, a new version of the transcript is created and stored.
AI Model Performance
In production, the system’s performance is continuously monitored in real time, with automated mechanisms in place to detect issues as they occur.
The AI model and its processes are well documented and generally understandable for most stakeholders, including business users, auditors, and regulators. However, further improvements could make explanations even clearer and more accessible.
The iBabs Debrief AI model is monitored through automated performance dashboards and periodic human reviews to ensure continued accuracy, explainability, and compliance with Euronext Corporate Solutions AI governance framework.
Version and Update Management
Current Software Version: iBabs Debrief v1.0
Azure Dependencies: Azure Cognitive Services (Speech API v3.1, Text Analytics v3.2)
Update Policy: Security patches are applied in accordance with Section “Information Security”.
Compatibility: Backward compatibility is ensured for all minor releases.
Considerations
Quotas and Limitations
Service Quotas and limits - https://learn.microsoft.com/en-us/azure/ai-services/speech-service/speech-services-quotas-and-limits
Monitoring and Uptime:
- The system monitoring is implemented.
- The system is designed to work 24/7 within the stipulated RTO and RPO.
User-side technical and hardware prerequisites:
- Supported browsers: Chrome 90+, Edge 90+, Safari 14+
- Internet connection: minimum 5 Mbps for uploading audio/video files
- Supported audio and video formats: MP3, MP4, WAV, M4A, AAC (up to 1.5Gb per file) and others
- Hosting: Azure Data Centers (EU/UK)
- Instances: Azure App Service (Standard tier)
Storage: Azure Storage with geo-redundant storage (GRS)
Information Security
The iBabs Debrief system is designed to ensure full compliance our NEN/ISO:27001/2 certification, actively mitigating risks associated with system vulnerabilities and safeguarding the integrity, availability, and confidentiality of data. Through robust security protocols, continuous monitoring, and adherence to best practices, the project reinforces a secure environment for sensitive information. These measures not only protect against potential threats but also uphold the trust and reliability expected within Euronext’s operational framework.
Microsoft Azure acts as a sub-processor, under a signed DPA, providing infrastructure and AI capabilities under standard contractual and security terms. Sub-processor performance and compliance are reviewed periodically. Under this arrangement, Microsoft processes data solely under the instructions of iBabs and in accordance with the contractual safeguards established by Euronext Corporate Solutions.
Data encryption
The system employs a layered encryption strategy to protect sensitive data both in storage and during transmission.
- Encryption in Transit: All data exchanges between services and clients occur over HTTPS (TLS), preventing unauthorized access or modification during communication.
- Encryption at Rest: Data is encrypted on both the server and client sides. Server-side encryption uses Azure service-managed keys, while client-side encryption utilizes customer-managed keys stored in Azure Key Vault.
This encryption framework aligns with industry best practices for data protection, ensuring compliance with enterprise security standards and regulatory requirements.
Artificial Intelligence Act (AI Act)
iBabs Debrief is classified as a limited-risk AI system under the EU AI Act.
The iBabs Debrief system is committed to full compliance with the Artificial Intelligence Act (AI Act), ensuring transparency, accountability, and ethical use of AI technologies. As an AI-powered solution designed to record and transcribe meetings, iBabs Debrief incorporates a range of measures to meet the AI Act’s obligations.
These include informing users when they are interacting with AI, tagging synthetic content, ensuring technical robustness and reliability, and implementing strong data protection aligned with GDPR. Users are required to obtain consent from meeting participants prior to transcription, and mechanisms are in place to allow access to, and withdrawal of, transcribed data. All AI-generated outputs are subject to user validation before finalization, and technical documentation—including performance testing, encryption protocols, and bias mitigation—is maintained to support auditability and compliance.
These efforts reflect Euronext Corporate Solutions proactive approach to responsible AI deployment and its alignment with European regulatory standards. Technical documentation, post-market monitoring, and transparency obligations are maintained throughout the system lifecycle to support accountability and regulatory readiness.
As iBabs Debrief operates entirely within Microsoft Azure, the solution benefits from Azure’s independently certified AI, security, and privacy frameworks. Microsoft’s documentation provides additional transparency on the design and governance of the underlying models and infrastructure that support iBabs Debrief, including but not exclusively:
- Microsoft AI Act - https://blogs.microsoft.com/on-the-issues/2025/01/15/innovating-in-line-with-the-european-unions-ai-act/
- Azure-wide security and privacy protection - https://azure.microsoft.com/en-us/blog/trusted-cloud-security-privacy-compliance-resiliency-and-ip/
Post-Marketing performance assessment as implemented by iBabs Debrief allows a continuous monitoring system for the performance of iBabs Debrief under operating conditions.
The monitoring objectives are (i) Verify that the system maintains its expected performance in production, (ii) Quickly identify any quality degradation or emerging risks, (iii) Detect unforeseen or problematic uses, (iv) Collect user feedback for continuous improvement, (v) Ensure ongoing compliance with AI Act and GDPR requirements.
Data Use and Compliance
Use of Pre-Trained Models Only
iBabs Debrief uses pre-trained language and speech models provided by Microsoft Azure.
iBabs Debrief does not train, fine-tune, or otherwise modify any AI models using customer or user data. All AI capabilities (such as speech-to-text transcription and content summarisation) rely on pre-existing Azure services that operate independently of customer datasets.
Microsoft Azure provides these models as part of its Cognitive Services suite as defined here. No customer data is used by Microsoft to retrain or fine-tune such models in accordance with the signed DPA.
Data Processing within Microsoft Azure
All processing activities take place entirely within the secure Microsoft Azure environment. Customer data (e.g. meeting audio, transcripts, or related information) is processed solely for the purpose of providing the agreed services — such as transcription, speaker identification, and summarisation. iBabs acts as a data processor on behalf of its customers, who act as data controllers and determine the lawful basis for processing (e.g., consent or legitimate interest) under Article 6 GDPR and UK GDPR.
Data processed through iBabs Debrief is not used for:
- Training, fine-tuning, or improving our or third-party language models;
- Sharing or enrichment of datasets; or
- Any purposes beyond service delivery.
No External AI or Third-Party Access
iBabs does not use or connect to any external AI systems or APIs outside of Microsoft Azure — including direct access to ChatGPT or other OpenAI systems. All AI processing and storage remain fully contained within Azure data centres located in the European Union or the United Kingdom. All processing within Azure data centres is subject to continuous compliance monitoring under ISO 27001, ISO 27701, and SOC 2 Type II certifications, ensuring data protection and operational resilience.
Compliance and Data Protection
The use of AI within iBabs Debrief fully complies with:
- The General Data Protection Regulation (GDPR);
- UK GDPR and Data Protection Act 2018;
- The Microsoft Azure Privacy and Security Framework; and
- Relevant European and UK AI governance principles (including human oversight and transparency).
All AI processing and storage remain fully contained within Microsoft Azure data centres located in the European Union and the United Kingdom. Any transfers of personal data between these regions occur under the UK–EU adequacy framework, ensuring equivalent levels of protection. iBabs and its parent company Euronext Corporate Solutions are committed to ensuring that all AI-based functionalities are transparent, secure, and ethically designed.
iBabs upholds the data protection rights of individuals under both the EU and UK GDPR, including rights of access, rectification, erasure, restriction, portability, and objection. Oversight is ensured under the supervision of the relevant data protection authorities, including the Dutch Autoriteit Persoonsgegevens and the UK Information Commissioner’s Office (ICO). Regular internal reviews and third-party audits are conducted to validate compliance with information security, privacy, and AI governance standards. Audit logs are retained for traceability, and any incidents are managed under Euronext Corporate Solutions Information Security Incident Response Procedure (ISIRP).
Data Protection Impact Assessments (DPIAs) are conducted where required to evaluate and mitigate potential risks associated with AI-based processing activities.
Summary
iBabs Debrief provides a secure, transparent, and efficient solution for meeting transcription and summarisation within the Microsoft Azure environment. The module has been developed according to Euronext Corporate Solutions privacy-by-design and responsible-AI principles, ensuring that all processing remains compliant with EU and UK data protection and AI governance standards. Customer data stays under the customer’s control at all times, supported by robust encryption, access management, and full data segregation. iBabs Debrief’s use of pre-trained AI technology within Microsoft Azure ensures secure, efficient, and compliant meeting transcription and summarisation without any use of customer data for model training or fine-tuning.