Bingli enables Schmitt-Thompson in both patient and nurse environments.

In collaboration with Schmitt-Thompson Clinical Content (STCC), we developed a clinically governed AI system designed to support nurses during live patient encounters without altering established clinical decision-making processes.

01
Supporting nurse advice line workflows.

Nurse advice lines are increasingly central to healthcare access, but rising patient volumes, staffing constraints, and documentation demands place significant pressure on clinical staff. Nurses must often navigate structured protocols, document interactions, and manage multiple systems while simultaneously conducting complex triage assessments.

Bingli’s ambient listening platform addresses these challenges by capturing real-time nurse–patient conversations, identifying relevant STCC protocols, and pre-populating structured questions based on information already provided during the call. All AI-generated suggestions remain traceable to the underlying transcript, enabling nurses to verify and validate outputs before incorporating them into clinical documentation.

02
Human-in-the-loop clinical design.

The system is explicitly designed as assistive technology rather than autonomous decision support. Nurses retain full responsibility for selecting protocols, prioritizing questions, interpreting patient responses, and making final triage and disposition decisions. AI is used solely to reduce administrative burden and improve information organization while preserving the integrity of established STCC clinical guidelines.

03
AI validation and performance benchmarking.

A dedicated validation platform underpins Bingli’s development approach. More than 4,400 simulated nurse–patient conversations have been generated and clinically validated across approximately 856 adult and pediatric STCC guidelines. This dataset is used to benchmark model performance across key dimensions such as protocol recommendation accuracy, question pre-fill correctness, hallucination rate, latency, and performance under varying conversational complexity.

This continuous evaluation framework enables systematic identification of underperforming areas, iterative model improvement, and transparent monitoring of system behavior in clinically sensitive workflows.

04
Responsible AI for nurse-led triage.

By combining STCC’s established evidence-based clinical protocols with a validated ambient AI platform, the partnership between Bingli and STCC aims to support healthcare organizations in modernizing nurse advice line operations. The focus is on improving efficiency, reducing administrative workload, and maintaining strong clinical governance and transparency.

4,400+ validated transcripts

These transcripts were created with controlled “noise” levels. All transcripts were reviewed by medical experts who validated the correct answers to be prefilled. Models were evaluated on accuracy, hallucination rate, latency, and cost.

+30% accuracy increase

+ 30% increase in guideline recommendation accuracy using Bingli’s optimized, AI-specific guideline definitions compared to native STCC definitions.

92-93% prefill accuracy

prefill accuracy reached by most models tested, with hallucination rates up to 12%. Noise levels of up to 70% irrelevant information had no impact on performance.

Integrating AI into Nurse Advice Line Services

Discover Smart Nurse Triage

Key strengths and benefits

4,400+ validated transcripts

+30% accuracy increase

92-93% prefill accuracy