Corti Assistant

​

New

• Added configureApp() and setInteractionOptions() to the Embedded API, splitting app-level configuration from interaction-level options such as mode, spoken language, template defaults, and document actions. The existing configure() and configureSession() methods are now deprecated and will remain supported during the deprecation period. See the configuration migration guide and the Embedded API deprecation timeline for more information
• Dictation: Expanded the previously introduced punctuation and voice command controls with in-session access, supporting easier control for end-users in an embedded context. Spoken punctuation and voice commands are now enabled by default, while remaining configurable by end users

​

Improvements

• Enhanced handling of deleted interactions to prevent archive errors when records are removed through retention policies or external APIs

​

Fixes

• Fixed an issue where Embedded API interaction-scoped methods did not always work correctly when Assistant was opened on an existing interaction

​

Improvements

• Improved UI translation support and resolved UX inconsistencies in the redesigned session configuration experience ahead of broader rollout

​

Improvements

• Session Configuration (Early Access): Introduced a redesigned session configuration experience, including improved microphone setup guidance for in-person ambient sessions to help improve transcript and documentation quality. The experience is currently rolling out progressively to selected customers ahead of broader availability
• Increased minimum iframe height requirement (480px) The embedded Assistant iframe now requires a minimum height of 480px (up from 320px) to support the new session configuration design, including its compact variant for smaller screen sizes. Please update your iframe configuration accordingly

​

Fixes

• Fixed an issue where getStatus() could return the most recently created interaction instead of the currently displayed interaction
• Rolled out a critical security update

​

Improvements

• Enabled diarization for ambient recordings to improve speaker separation in generated transcripts while preserving the existing transcript view without speaker labels
• Added support for hosting Assistant under a configurable non-root path (e.g. /corti-assistant/), ensuring all application routes and assets resolve correctly within the configured proxy prefix. For more information, read our proxy guide

​

Fixes

• Fixed an issue where configured websocketBaseUrl values were not consistently respected and could fall back to the default URL
• Rolled out a critical security update

​

New

• Template Customization: Concluding the progressive rollout of the Template Assembler, the feature is now enabled by default in Assistant and includes the latest customization options for adjusting section writing style, formatting, and adding instructions. If needed, end-user access can still be disabled via the Embedded API configuration. Standalone customers can contact help@corti.ai to disable it
• Dictation (Beta): Added support for controlling punctuation and voice commands via the UI. From the Input tab in Settings, users can choose whether spoken punctuation (e.g., “period”, “slash”) is converted into symbols, and whether voice commands are enabled for more hands-free navigation. Easier access in embedded contexts is coming in a future release

​

Improvements

• Added guardrails to prevent silent failures when selecting unsupported dictation languages via embedded configuration
• Introduced grouped documentation sections in the session UI using the Group Name :: Section Name convention, making it easier to copy related sections mapped to the same EHR field
• Added a visual indication when the end-user triggers sending documentation to the host application, to confirm the action and prevent duplicate submissions

​

Fixes

• Fixed an issue where parts of the generated documentation would sometimes appear in an incorrect language
• Resolved an issue where the back button would navigate within the iframe instead of the parent page
• Fixed an issue where the new session screen displayed the wrong session type on first load in Safari

​

New

• Web Component API Now Recommended: Documented the Web Component API as the new preferred integration method for Corti Assistant. This approach provides a cleaner, more maintainable integration pattern compared to postMessage or Window APIs.
• New React Integration Guide: Added a comprehensive React integration guide demonstrating best practices for embedding Corti Assistant in React applications using the Web Component API.
• Improved Navigation and Entry Points: Restructured Assistant documentation with a new welcome page and integrations overview to help developers quickly find the right integration approach for their needs.

​

Fixes

• Fixed an issue where the “Template Management” label in the template picker did not follow the selected UI language
• Fixed an issue where dictated text could appear duplicated in AI Chat
• Fixed an issue where some sessions did not automatically receive a generated title

​

Fixes

• Fixed an issue where the shortcut to generate the default document would intermittently not appear

​

Improvements

• Internal maintenance updates

​

Improvements

• Internal maintenance updates

​

Fixes

• Fixed inconsistencies in postMessage response handling documentation
• Fixed an issue that made switching between user accounts more difficult

​

Improvements

• Improved handling of built-in templates to ensure only correctly configured templates can be generated
• Expanded availability of Template Assembler by adding additional breakpoint thresholds, enabling support for a broader range of use cases
• Added required clipboard-write permission to the iframe to enable copying documentation

​

Improvements

• Re-enabled interim text during dictation for languages where supported. Learn more here

​

Fixes

• Continued to consistently send the deprecated ready event on every authentication request, to ensure backwards compatibility

​

Improvements

• Simplified microphone selection by centralizing microphone selection in session only
• Updated microphone selection to remember the last-used device via local storage when no user preference is set

​

Improvements

• Internal maintenance updates

​

Improvements

• Internal maintenance updates

​

New

• Added getTemplates method to the Embedded API to retrieve available document templates for the authenticated user

​

Improvements

• Improved validation to prevent document generation when context is empty (no facts or notes present)
• Enhanced Embedded API error handling with clearer, human-readable validation messages for all API methods
• Updated configureSession documentation to clarify input restrictions, improved error handling, and specified behavior in case of user defaults (no overwrite)
• Optimized application loading for faster initial state rendering

​

Fixes

• Fixed an issue where ending dictation in context notes cleared the note content
• Fixed all dictation buttons becoming active simultaneously on mobile devices
• Fixed note content loss when navigating to newly created interactions during rapid input

​

Fixes:

• Fixed an issue where Template Assembler did not show up in standalone even when enabled

​

New

• Template Customization: Introduced the Template Assembler, allowing end-users to customize documentation templates by adding, removing, reordering, and renaming template sections, as well as editing the template title. The feature is currently hidden by default behind a feature flag and will be rolled out progressively over the next two months. Customers interested in learning more can reach out at help@corti.ai.
• Introduced a new embedded event system with many more events to subscribe to and a lot more detail per event. The legacy event system is deprecated, but still supported.
• Norwegian Nynorsk and Norwegian Bokmål now available as selectable UI languages

​

Improvements

• Removed the success toast message that appeared after pressing the Synchronize Document button
• Enabled users to access the Medical Device instance of the product via Mobile Applications

​

Fixes

• Fixed a bug that prevented templates in the default template’s language from being prioritized in the template picker
• Fixed a bug that prevented the template picker to display under some circumstances

​

Improvements:

• Improved data fetching performance

​

Fixes:

• Fixed an issue where document sections sometimes switched order

​

Improvements:

• Added stickiness to section headings and toolbar in documents so they remain visible while scrolling, including background to make sure there is no text overlap.

​

Fixes:

• Fixed an issue that was preventing changing language and setting overrides at the same time using the Embedded API method configure
• Fixed a bug where it was not possible to continue dictating after stopping the dictation recording
• Fixed an issue that led template sections to appear in english by default
• Fixed a bug that was rendering the template picker empty when the page was resized beyond a certain screen resolution during usage

​

Fixes:

• Fixed issue causing certain public API endpoints to be unusable

​

Improvements:

• Added caching to improve application loading speed

​

Fixes:

• Fixed some issues that were rendering dictation to malfunction in the context tab

​

Fixes:

• Fixed a bug where dictation language could not be set
• Fixed a bug where document payloads were not aligned between different events

​

Improvements:

• Added the ability to configure the presence and label of a “Synchronize Document” button for embedded flows

​

Fixes:

• Fixed a bug where application unintentionally discarded recently dictated notes

​

Fixes:

• Rolled out a critical security update

​

Fixes:

• Rolled out a critical security update

​

Fixes:

• Rolled out a critical security update

​

Improvements:

• Added caching to improve application loading speed
• Updated MediaRecorder settings to provide more optimal audio configuration for ambient scribing
• Enabled users to change their default template directly from the template picker

​

Fixes:

• Rolled out a critical security update

​

Fixes:

• Fixed an occasional mix-up in speaker identification within the transcript, during virtual sessions

​

Fixes:

• Fixed an issue where the default template did not appear in the template picker

​

New:

• Added the ability to configure the application interface through the Embedded API
• Added support for setting credentials via the Embedded API

​

Improvements:

• Improved the authentication flow with a new getStatus method to retrieve the application’s current state, including authentication status, user info, URL, and interaction details
• Enhanced the template picker by sorting templates alphabetically within each language
• Updated the template picker to prioritize templates that match the language of the active user’s default template

​

Fixes:

• Fixed reloading the window on re-authentication attempt when the session was already initialized

​

Fixes:

• Resolved a configuration error that caused German dictation to malfunction

​

Improvements:

• Facts headings now consistently display in the user’s chosen UI language, ensuring a clearer and more predictable experience for those working across multiple languages

​

Fixes:

• Fixed Fact headings in new sessions rendering in a code-like format in the Context tab
• Fixed Fact headings occasionally appearing in the user’s default spoken language

​

Improvements:

• Added temporal grouping in transcripts, in languages that previously didn’t support it
• The microphone volume indicator has returned with an improved, WCAG-aligned design to support clearer and more accessible audio setup

​

Fixes:

• Fixed scrolling behavior to ensure documents scroll correctly again
• Resolved an issue where session language defaulted to English in embedded mode

​

Improvements:

• Disabled the sound associated with muted-microphone alerts to reduce disruptive prominence in specific setups

​

Improvements:

• Lowered the threshold for silence detection in streams, to reduce the sensitivity of muted-microphone alerts

​

New:

• Dutch and Norwegian available as selectable dictation language

​

Improvements:

• Added a microphone mute alert to notify users when their microphone is muted
• Added support for selecting the current tab as a virtual input source when using the embedded applications

​

Fixes:

• Fixed an issue where requests to the AI Chat were rejected due to consultations not being registered as ended
• Fixed a race condition that prevented the ?language= parameter from reliably setting the session language