AI knowledge assistant for healthcare contact centres
Empowering contact centre agents with real-time medical protocol retrieval and automated call transcription.
Star-shl
Star-shl, a leading Dutch healthcare diagnostic centre, needed to reduce the onboarding time for new contact centre staff and ensure consistent adherence to complex medical protocols. The goal was to build an AI-powered assistant that could listen to live calls, transcribe them, and instantly provide agents with the correct procedural information from their extensive knowledge base.
Contact centre agents at Star-shl manage a vast array of medical protocols, from blood collection procedures to DVT scan requirements. Manually searching through Confluence and internal documents during a live call is slow and error-prone. Furthermore, ensuring that agents do not inadvertently share restricted medical results over the phone is a critical compliance requirement that needed automated monitoring.
A look inside
What SevenLab built
SevenLab developed a custom AI agent that integrates real-time speech-to-text with a sophisticated RAG (Retrieval-Augmented Generation) system. The platform transcribes the agent's side of the conversation using the Mistral Large model and instantly queries a vector database containing over 190 medical protocols to provide immediate, actionable advice.
Real-time transcription
Live audio streaming and transcription of agent speech using sovereign European AI models.
Protocol retrieval
Automated document matching that surfaces the exact medical SOP or FAQ needed for the current context.
Compliance monitoring
AI-driven alerts that prevent agents from sharing medical diagnoses or results in violation of protocols.
Multi-source integration
Seamless synchronisation with Confluence and internal HTML/PDF knowledge bases for up-to-date info.
Measurable business impact
The AI assistant transformed the contact centre workflow by providing instant answers to complex queries, such as DVT scheduling rules. It successfully flagged compliance risks in real-time and provided direct links to source documentation, ensuring agents could verify information without leaving the chat interface.
The interface looks beautiful and simple. Seeing how it correctly references our SOPs and prevents us from sharing sensitive diagnoses is exactly the solution we need for our team.
Want results
like these?
Tell us your challenge and we'll show you how we'd solve it — with a clear scope, timeline, and fixed price.
Talk directly with our AI specialists