Skip to main content
This release adds Embedded Forms for the Ambient APIs (v1.7.0) and the Headless Web SDK (v0.3.1) so you can send the custom fields from your existing note templates on an ambient session and read filled values after the note is ready, Content Rewrite in the hosted Dictation SDK UI (v0.1.0-beta.3) so clinicians can reshape selected text in place, and Content Rewrite for Dictation APIs (v1.7.0) so your application can rewrite a selected span and write the result back in your own UI.

​
What’s new

​
New Features

​
New features

Headless Web SDK v0.3.1Embedded Forms: You can now send the custom fields already embedded in your note templates, plus optional prior section text, on the ambient session you run in Headless. Seed with setSessionContext or updateSessionContext using sections[].precharted_content. For mid-visit updates, pass updateMask such as ["sections"] or "sections", and replace the full sections list in that call. After the note is ready, read filled values from section_structured_values with getAmbientSessionSectionStructuredData, getAmbientSessionStructuredData, the matching encounter helpers, or React hooks such as useUpdateAmbientSessionContext and useAmbientSessionSectionStructuredData. Filled values use the shape { id?, context?, key, value }.
  • Your app decides when to refetch. The SDK does not poll in the background.
  • Note structured data still returns diagnoses and orders only in this release.
  • Headed Web SDK UI and Form filling are out of scope for V1.
  • Prefer setSessionContext first when you need offline upload of context.
Why it matters: Port the custom fields in your note templates onto the Headless ambient session you already run, fill them from the visit, and map discrete values back into your UI without mixing Form filling contracts.Learn more in the Embedded Forms overview guide.
API v1.7.0Embedded Forms for Ambient APIs: You can now send the fill-in-the-blank fields from your existing note templates with an ambient session, plus optional prior section text, and read the answers back as separate values for voice-driven fills in partner UIs. Add precharted_content to each section in session context, with type: "structured" entries for templates and type: "string" entries for prior section text. After the note is ready, read the filled values from the new section structured data endpoints or from section_structured_values in session and encounter structured data.Why it matters: Keep the custom fields already in your note templates, fill them from the ambient visit, and write discrete values straight into those fields so clinicians do not retype answers they already gave.Learn more in the Seed and read Embedded Forms guide. For how templates work, refer to Embedded Forms.
Dictation SDK v0.1.0-beta.3Content Rewrite in hosted Dictation: Clinicians can now rewrite selected text in the hosted Dictation editor when backend enablement allows it. With the microphone off and at least two words selected, they choose Concise, Verbose, or Patient Friendly. The UI shows Rewriting…, then replaces the selection on success, or shows Rewrite failed. Please try again. so the clinician can retry.
  • Partner apps keep the existing ShowOptions and onSubmit / onDraft / onCancel contract.
  • There are no new SDK methods or callbacks, and there is no new package to install. Keep v0.1.0-beta.3.
  • Web SDK Dictation uses the same iframe. The SDK does not expose an org or user switch for enablement.
Why it matters: Clinicians can reshape selected Dictation text in place without you wiring a rewrite API or changing your Dictation integration.Learn more in the Content Rewrite guide.
API v1.7.0Content Rewrite for Dictation APIs: You can now rewrite a selected span of clinical text after Dictation, or any plain text your application owns. Send the selection, surrounding context, and one instruction (CONCISE, VERBOSE, or PATIENT_FRIENDLY). The API returns the rewritten text and selection offsets so your UI can replace the span. Your application owns apply and save.
  • POST /api/v1/content/rewrite.
  • Keep selected_text and surrounding_content within 5,000 Unicode code points each. Free-form custom instructions are not supported.
Why it matters: Providers can reshape wording that is already in the chart without retyping or starting another Dictation session, while your application keeps control of selection, write-back, and save.Learn more in the Rewrite Dictation content guide. For the full API reference, refer to Rewrite content.
Last modified on October 16, 2026