Scenario conversations
AI may help create realistic conversation practice for everyday speaking situations assigned or reviewed by a therapist.
AI Ethics
AI that supports communication — not replaces care.
EaseTalk uses AI to support structured talking activities, reflection, and clinician visibility. AI does not diagnose, make treatment decisions, or replace Speech and Language Therapists.
AI should reduce friction between sessions, not remove human care. EaseTalk is designed to keep clinicians, clients, and privacy at the centre of the experience.
Our AI position
Helpful support, clear limits, human judgement.
EaseTalk uses AI as an assistive layer around therapist-supported communication work. It can help create practice context, reduce admin friction, and make between-session activity easier to review. It is not a clinician, a diagnostic system, or a substitute for professional care.
Where AI may be used
Designed for support around talking activities.
Depending on the feature and setting, AI may be used in the following ways. Some capabilities may be experimental, limited, or introduced gradually.
Speech-to-text
Speech technology may help convert spoken practice into text for review, reflection, or clinician visibility.
Text-to-speech
Generated speech may support guided prompts, scenario flow, or accessibility in talking activities.
Reflection prompts
AI may help users reflect on confidence, difficulty, and what felt important after a practice activity.
Progress summaries
AI may help summarise engagement signals so clinicians can review between-session activity more efficiently.
Clinician workflow support
AI may help reduce repetitive admin around reviewing activities, while leaving clinical interpretation to the SLT.
Clear boundaries
What AI does not do
These limits are central to how EaseTalk should be understood and used.
- No diagnosisEaseTalk does not diagnose speech, language, communication, mental health, or medical conditions.
- No independent treatment decisionsAI does not decide therapy plans, goals, or clinical next steps.
- No replacement for SLTsEaseTalk supports between-session work and clinician visibility; it does not replace Speech and Language Therapists.
- No emergency or crisis supportEaseTalk is not designed for urgent, emergency, or crisis situations.
- No fully automated clinical decisionsAI outputs should be reviewable and should not be treated as final clinical judgement.
Human oversight
SLTs remain central.
EaseTalk is designed to support the relationship between a client and their Speech and Language Therapist, not step around it.
Reviewable outputs
AI outputs should be visible, editable, and possible to disregard where they are not useful or appropriate.
Clinical interpretation stays human
Clinicians remain responsible for clinical interpretation, care decisions, and how information is used in sessions.
Between-session context
EaseTalk helps surface activity, confidence, and reflection signals so sessions can begin with clearer context.
Client control
Users and clinicians should retain control over what is reviewed, corrected, shared, and acted on.
Data and privacy
Privacy-first by design.
Speech, transcripts, reflection notes, and progress data can be sensitive. EaseTalk treats that sensitivity as a product requirement.
Data minimisation
EaseTalk is designed around collecting and retaining only what is needed to provide the service and support clinical review.
EU-region infrastructure
Platform data is handled using EU-region infrastructure where appropriate for the service.
No training on identifiable user data
Identifiable user data is not used to train AI models.
Speech and transcript care
Speech and transcript handling should be privacy-first, access-controlled, and proportionate to the feature being used.
Bias, speech difference, and accessibility
Speech technology can be imperfect.
AI and speech systems can misunderstand people. EaseTalk should not treat AI outputs as perfect, especially where communication difference is part of the user experience.
Speech difference matters
Outputs may be affected by dysarthria, stuttering, aphasia, voice quality, pace, fatigue, or other communication differences.
Context matters
Accents, language variation, background noise, microphone quality, and device quality can affect system performance.
Control matters
Users and clinicians should be able to correct, ignore, or challenge AI outputs that do not reflect the person accurately.
Governance and future commitment
We expect our responsibilities to keep evolving.
EaseTalk will continue reviewing AI features with clinical input, user feedback, privacy obligations, and evolving regulation. As new capabilities are introduced, we will keep assessing what they are for, what they should not do, and how people remain in control.
Read the detailed policies
For legal terms, data handling details, or questions about EaseTalk, use the links below.