AI-driven language and communication skills training for social carers

Around 300,000 non-EU staff work in direct care roles across the UK. To have secured a Health and Care Worker visa, an overseas applicant must have demonstrated a B1 level of English, an intermediate level slightly above GCSE. However, this level often is not sufficient for the complex communication demands of care work. As a result, there are many care workers who do vital work but struggle with the English language skills needed to manage the challenging communication demands such as navigating regional accents, understanding idiomatic expressions, and holding sensitive conversations with confused, distressed, or grieving service users and their families.

This project will develop an AI-powered language tool to build carers’ confidence and communication skills. Accessible via web browser and mobile app, the tool will use real-time AI-driven roleplay and feedback to simulate natural care-related conversations and provide contextual learning. Powered by a Large Language Model (LLM) with Retrieval-Augmented Generation (RAG), the AI will dynamically reference relevant content to offer accurate, safe, and situation-specific guidance. Reinforcement learning from human feedback will be included to refine responses and provide contextually accurate, ethical, and safe advice tailored to carers’ real-world communication needs.

Designed with carers’ mobile-first preferences in mind, the tool offers personalised, on-demand support aligned with real-world needs. With strong sector relationships, SLC has identified a clear appetite for affordable, accessible language training—filling a critical gap and supporting better care outcomes across the UK.