🚀 Overview
In this release, we’ve focused on improving the underlying architecture of Ambient Scribe to ensure faster, more resilient transcription and document generation. From upgraded infrastructure and services to better frontend stability and smarter AI outputs, these enhancements reduce friction and optimize the overall clinician experience.
🌇 Voice Profile Feature Sunset
As part of our transcription improvements, we’ve sunset the voice profile registration feature. This change does not impact documentation quality or functionality. Instead of assigning names to speakers, diarization will now use clear, numbered labels like Speaker 1, Speaker 2, etc. The updated approach improves reliability and avoids confusion, especially in multi-speaker sessions. You’ll continue to benefit from diarization — now with fewer mismatches and no additional setup required.
🔍 What’s New?
✨ Enhancements:
- AI Disclaimer Update (MED-2860): The AI-generated disclaimer text now clearly states “AvodahMed Ambient Scribe” across all locations (on-screen, PDFs, and EHR notes).
- Appointments Page Redesign (MED-2410): The Appointments page has been updated with a cleaner layout and better usability based on new Figma designs.
- Improved Troubleshooting Logs (MED-2866): Document metadata component unmasked in DD for better diagnostic visibility.
- Efficiency Gains (MED-2644): Frontend polling replaced with a pub/sub system to improve Nsight performance and reduce Aurora costs.
- Transcription Safety Logic (MED-3014): Added safeguards to prevent data loss from malformed transcription provider flags.
- Symptom Tracking: Vital Spine & Pain (MED-2834): Enhanced detection logic for this customer protocol set.
🐞 Bug Fixes:
- Send to EHR UI Fix (MED-2873): Cleaned up vertical stacking and missing button styling in the documentation editor.
- DocWizard Template Saving (MED-2677): Custom template edits now persist correctly.
- Appointments History Accuracy (MED-2845): Fixed method filter miscounts on the History page.
- Filter Isolation (MED-2972): Filters now behave independently across pages.
- Custom Workflow Stability (MED-2962): Fixed CRUD issues in workflow configuration.
- Live Timer Accuracy (MED-2779, MED-3009): Timer no longer resets during websocket connection changes.
- Quick Start Dashboard Issue (MED-2966): Resolved provider display issues for completed appointments.
- Dictation Resilience (MED-2948, MED-2898): Dictation no longer randomly stops or cuts off — especially on smaller screens.
- Athena Ingestion Fixes (MED-2958): Improved appointment sync handling for VEHR/Athena.
- UI & Error Messaging (MED-2964, MED-2878, MED-2895, MED-2938, MED-3005): Addressed multiple bugs related to error messages, ICD tooltip behavior, cache clearing, and document save indicators.
⚙️ Other Improvements:
- Transcriber Audio Recorder (MED-2963): Our live dictation editor now uses the transcriber for more reliable audio capture.
- Stream Reliability Boost (MED-2904): Retry logic added to Kinesis streams to better distribute load and reduce throughput errors.
- Redis Migration (MED-2789): Switched to AWS-managed Redis for better stability and scalability.
- Error Messaging Inventory (MED-2786): Centralized mapping of error messages and logic triggers in Nsight.
- Monitoring Upgrades (MED-2992, MED-2970, MED-2742, MED-2997, MED-2882, MED-2668): Multiple improvements to logging, source mapping, service instrumentation, and Lambda migrations.
- Diarization Pub/Sub Events (MED-2853): Diarization service now emits completion events to reduce delays in downstream processing.
- Live Dictation Enhancements (MED-2971): Users now see non-final words as they speak — a visual indicator that dictation is actively capturing input.
- Feature Flag Flexibility (MED-2986, MED-3007): AWS Transcribe and Medical Transcribe support improved for both live and diarized audio sessions.
📚 Need Help?
For more information, visit our KnowledgeHub or reach out to our support team at support@avodah.com.