EZEval

This tool is restricted to authorized clerkship faculty. Enter the access password shared with you by your program director.

🔒

FERPA: This app holds the draft in browser memory and clears it after inactivity. Student names, observations, and conversation history go to the configured inference service (Anthropic, Google, or an Ollama endpoint). Use invented details for a demo unless your institution has approved this workflow.

🚫

PHI: Microphone audio is processed on-device only. Do not include patient identifiers unless your institution has approved this workflow; use clinical descriptors such as "a patient with hypertension."

🤖

AI and speech: Speaking mode requires verified on-device speech recognition. Typing mode works without a microphone. The model adapter receives transcript text and returns evaluation guidance.

⏱

Auto-Clear: All session data clears after 15 minutes of inactivity or when you end the evaluation.

⚕

Welcome to the FM Clerkship Evaluator

An AI assistant will guide you through all 8 competencies, ask follow-ups, infer grades, and draft feedback for MedHub.

Choose Typing or Speaking above. Speaking requires on-device speech recognition and microphone access.

Invented observations · model-inferred grades · no microphone

⚕

Ready

⏸ Conversation is currently paused

🤖 AI-powered · 🔒 No data stored · Use Chrome for voice

⚠️ DRAFT — FOR MEDHUB ENTRY ONLY — Do not save, print, email, or forward. Clear clipboard after pasting.

Competency Grades AI-recommended · radio buttons editable before MedHub entry

Competency	Acquiring Fail	Appropriate Pass	Exceeding Near Honors	Exceptional Honors	Not Observed

Formative Feedback AI-drafted from your observations · click Copy to paste into MedHub

Overall Summary

—

Inspect transcript and model decisions

The evidence links connect the app's eight competency descriptions to model-cited phrases found verbatim in the preceptor's spoken or typed observations. They are citations for review, not a view of model attention or proof that the grade is correct. The institution's full rubric is not loaded here. Every model request also includes prior conversation history and the server-side policy. Audio and hidden model reasoning are not recorded. This view stays in browser memory and clears with the evaluation; share identifiable content only through an approved research workflow.

Competency evidence and model judgments

Turn-by-turn record

Welcome to the FM Clerkship Evaluator

Review Results — —