Voice AI in Healthcare: How Marra Makes Technology Accessible to Every Elderly Person

When we talk about AI in healthcare, the conversation often centres on cutting-edge diagnostics, clinical decision support, or hospital workflow automation. These are important. But there is another frontier — one that is arguably more urgent and certainly more overlooked: making AI genuinely accessible to the elderly people who have the most to gain from it.

Marra's voice-first architecture is not a feature — it is a philosophy. Here is why.

The Accessibility Imperative

Consider what it means to interact with most health apps. You need a smartphone. You need to download the app. You need to create an account with an email address and password. You need to navigate menus and input text. You need a reliable internet connection.

For an 82-year-old who has a landline and a basic mobile phone, this is an insurmountable barrier. The technology literally does not reach them.

Voice as a Universal Interface

Every person who can have a conversation can use Marra. The AI calls them. They answer. They speak. That is the entire interface. No learning curve. No setup. No frustration.

The voice engine uses OpenAI Whisper for speech recognition, ElevenLabs for a warm, natural voice output, and Claude API for conversational intelligence. The system is designed with a deliberately slow pace, clear enunciation, and warm tone — tested specifically with elderly users rather than defaulting to settings optimised for younger demographics.

Multi-Language as Equity

1.5 million people in the UK speak little or no English. Many of them are elderly. Many have complex health needs. Most health technology is English-only.

Marra launches with English and adds Welsh, Urdu, Bengali/Sylheti, Punjabi, and Polish in Phase 2. Each language version is reviewed by native speakers and community health workers — not just machine-translated.

Accessible healthcare technology is not a nice-to-have. It is a matter of health equity.