Artificial intelligence (AI) continues to reshape healthcare, with recent UK guidance – “Guidance on the use of AI-enabled ambient scribing products in health and care settings” – encouraging the safe and effective adoption of AI-powered documentation tools. This reflects growing recognition that well-designed AI can help reduce clinical admin, improve documentation quality, and free up more time for direct patient care.

At G2 Speech, we’ve developed Aida, our AI assistant, to support healthcare professionals in producing clear, accurate, and structured documentation. Built into our platforms such as SpeechCursor and SpeechAmbient, Aida is tailored for clinical use – but we understand that many users have valid concerns around data security, privacy, and regulatory compliance.

In this FAQ, we’ll address some of the most common questions around how AI is being used in our products, and how we ensure it meets the highest standards of protection and governance in healthcare environments.

General Questions

What is Aida?  

Aida is G2 Speech’s AI-powered digital assistant, designed specifically to support healthcare professionals with medical documentation. It helps improve the clarity, structure and efficiency of clinical reporting, ultimately supporting better communication and patient care. 

How does Aida work? 

Aida uses Large Language Models (LLMs) and deep learning to transform dictated text into well-structured medical documents. It offers intelligent text optimisation, grammar correction, summarisation and translation – all embedded within our clinical dictation platforms like SpeechCursor, SpeechAmbient and soon SpeechReport later this year. 

What are the key features of Aida?  

• AI Assistant tools: Options like grammar correction, sentence restructuring, and document summarisation 

• Translation support: Multilingual output to improve inclusivity and accessibility 

• Rigorous Clinical Safety: AI safeguards and protections built-in to eliminate hallucination, inference, and unintended clinical decision making, supporting our Clinical Safety DCB0129 compliance 

• Privacy and security: Hosted in a UK-based ISO 27001-certified environment,  compliant with DSPT, and designed in line with UK GDPR and ISO 27701. Data is encrypted in transit and at rest, and no PHI is stored or used beyond the session

What are the benefits of Aida?  

• Improves clinical documentation clarity and consistency 

• Saves time by reducing manual editing and admin 

• Reduces cognitive load for busy clinicians 

• Enables professionals to focus on direct patient care 

• Integrates seamlessly with existing dictation workflows 

Technology and security

What Large Language Model (LLM) is used? 

G2 Speech’s AI assistant (Aida) uses Large Language Models (LLMs) integrated into our cloud platforms, hosted securely in the UK Azure region. We primarily work with trusted European and open-source providers, and continuously evaluate alternatives to ensure performance, privacy, and regulatory alignment. 

When using AI features (e.g. summarisation, grammar correction), only the dictated text and user prompt are sent to the model for real-time processing. No audio or recognised text is stored after the session, and no training of the model occurs using user data.  

Is patient information secure? 

Yes – all AI functions are delivered within our DSPT-compliant, ISO 27001-certified, ISO/IEC 27701-aligned, and Information Security and Privacy Management System. Specific safeguards include: 

• Data minimisation: Only the requested text and transformation are processed 

• Session-based operation: Data is not stored post-processing 

• Encryption: All data is encrypted in transit and at rest 

• Access controls: User identity and access are strictly managed within the G2 Cloud 

While dictated content may contain health-related terms, users are generally advised not to include personally identifiable information in AI prompts.   

How do you ensure security and compliance within the platform? 

Aida is designed to operate within a robust governance framework that meets the security and privacy expectations of healthcare professionals. Our platforms are developed and hosted in compliance with: 

• ISO/IEC 27001 (Information Security Management) 

• ISO/IEC 27701 (Privacy Information Management) 

• NHS DSPT and DCB0129 clinical safety standards 

• UK GDPR, including principles of lawfulness, purpose limitation, and data minimisation 

Whilst minimal, all personal data is processed strictly under the documented instructions of the data controller (Customer/User/NHS Trust), and only by personnel bound by formal confidentiality obligations. G2 Speech act only as a data processor. 

Security measures include: 

• Encryption in transit and at rest (TLS 1.2+ / AES-256) 

• Role-based access control (RBAC) to restrict access to authorised users 

• Multi-factor authentication for administrator access 

• Regular internal and external audits 

• Certified hosting within Microsoft Azure (UK region) 

What happens to any recorded data? Is it stored? 

 No. Aida is explicitly designed to minimise data retention and operate on a session-based model: 

• Dictated audio, recognised text, AI prompts, and AI outputs are not stored 

• No content is used to train AI models 

• No patient-identifiable data is retained beyond session duration 

Audio, transcribed content, AI inputs and outputs are processed in real time and discarded immediately after the response is returned. The only information retained is basic user account metadata (e.g. email address, registration status) – this is required to support authentication and usage monitoring. 

This design ensures compliance with GDPR Article 5(1)(e) (storage limitation), Article 32 (security of processing), and aligns with clinical safety requirements under DCB0129. 

Do you provide documentation for IG and clinical governance? 

Yes. We can provide governance documentation to support your internal assurance processes, including: 

• Data Protection Impact Assessments (DPIAs) 

• Data Processing Agreements (DPAs) 

• DTAC compliance summaries 

• Clinical Safety documentation (in line with DCB0129) 

• Certificates for ISO 27001, ISO 27701 and Cyber Essentials Plus 

Implementation and support

What does implementation look like? How long would it take to be live with the platform?  

Deployment is straightforward. G2 Speech provides user accounts and secure access to the platform via our user portal. ICT departments can install the lightweight desktop application (e.g. SpeechCursor) or enable access to the browser-based interface (e.g. SpeechAmbient). Internal testing and onboarding can typically be completed within a few days, depending on local approval processes and user training needs. 

As part of onboarding, we can also provide governance documentation (DPIA, DPA, DTAC, clinical safety case) to support Trust IG and digital approval processes. 

Do I need patient consent? 

No.  

Under UK GDPR, explicit patient consent is not required to use AI-assisted documentation tools like Aida, provided the processing is carried out under a valid lawful basis – typically Article 6(1)(e) (public task) or Article 6(1)(c) (legal obligation), with Article 9(2)(h) for processing health data for direct care. 

In this context, your Trust is the data controller and G2 Speech acts solely as a data processor, operating under the controller’s documented instructions. The Trust remains responsible for ensuring a lawful basis exists for all data processed through the platform. 

How will Aida integrate with existing platforms? 

For Speech Cursor (desktop dictation):  

Aida operates as part of the SpeechCursor environment and integrates seamlessly with third-party applications via: 

• Voice-driven text input into any active text field 

• Voice commands to trigger application-specific buttons 

• Simulation of keyboard shortcuts to interact with third-party systems 

These integrations require minimal setup and work with most EPRs, PAS systems, and clinical portals used in NHS and European healthcare settings. 

Integrations with third party applications can be set up easily, and can cooperate with: 

• Sending text to specific text fields using voice commands 

• Toggling buttons using voice commands  

• Triggering third party application keyboard commands  

For SpeechAmbient (browser-based, multi-speaker): 

SpeechAmbient is typically aligned with a customer’s existing EPR or documentation platform. Templates can be mapped to match local clinical formats, and structured summaries are processed into designated text fields or copy-paste interfaces, depending on the workflow. Integration does not require deep system-level access. 

Where is Aida already in use?  

Aida has been rolled out or piloted in multiple healthcare environments across the UK, Ireland, Belgium, France, and the Netherlands, supporting clinicians in areas such as pathology, neurology, community care, and a virtually any other specialty. 

Uptake varies by product (e.g. SpeechCursor, SpeechAmbient) and deployment model, but the AI functionality is now actively used in both live and trial deployments across several G2 Speech customer sites. 

Is it possible to trial SpeechCursor and/or SpeechAmbient with the AI digital assistant, Aida? 

Yes! Aida is available to be tested for two weeks (with up to five testers) completely free of charge. Request a free trial! 

What does customer support look like? 

We recognise that implementing new technology in clinical settings requires responsive, hands-on support. That’s why our team works closely with both clinical and IT teams throughout onboarding and beyond. 

Support includes: 

• Dedicated onboarding specialists to assist with installation, configuration, and first-line training 

• Access to our technical support team for issue resolution and usage queries (availability varies by region and contract tier) 

• System monitoring and health checks for G2 Cloud deployments 

• Ongoing training and documentation to ensure continued confidence and product adoption 

Customers with enhanced support agreements may also receive access to priority assistance, out-of-hours escalation options, and regular usage reviews. 

We understand that introducing new technology into healthcare workflows can be challenging, which is why our team is committed to providing seamless product integration and implementation from day one. Our dedicated onboarding specialists work closely with your clinical and IT teams to ensure minimal disruption and maximum efficiency. Beyond setup, our premium after-sales support includes 24/7 access to technical experts, regular system health checks, and ongoing training resources to keep your staff confident and your system performing at its best. 

What training packages are available?  

We provide a comprehensive e-learning platform for self-paced training, available in English, French, and Dutch. The content includes: 

• Core system functionality 

• Dictation workflows 

• AI assistant features 

• Tips for clinical efficiency and accuracy 

This supports both onboarding of new users and refresher training for existing staff. Additional in-person or virtual training sessions can be arranged upon request, depending on project needs. 

In development

How do you involve users in the development of products like these? 

We actively involve clinical users in the development and refinement of our products through a combination of:  

• On-site floorwalking during rollout phases, coordinated with your ICT team for live support and feedback 

• Ongoing user feedback channels, including structured surveys, early access programmes, and direct feedback via the G2 Service Team 

• Clinical engagement workshops to ensure features are aligned with real-world workflows and safety considerations 

All feedback is reviewed by our product and clinical safety teams, helping to shape future updates in line with both user needs and regulatory requirements (e.g. DCB0129, GDPR, ISO 27001/27701). 

What can we expect in the future roadmap?  

While specific roadmap items are still being prioritised, our current focus includes: 

• Enhanced integration capabilities with third-party platforms (e.g. EPRs, clinical systems) 

• New AI features to support more efficient structuring and summarisation of clinical content 

• Ongoing improvements to user experience and accessibility across all platforms 

We remain committed to developing features that reduce documentation burden and support safe, effective care delivery. 

Have More Questions? 

If we didn’t cover your question here, feel free to reach out at info@g2speech.com or visit the Aida page:  

https://www.g2speech.com/solutions/core-products/aida/