Amigo API

Release history for Amigo backend APIs, including Classic, Platform, and Scribe changes.

Release history for Amigo backend APIs. Latest numbered Amigo API release: v0.9.557 (July 2026). Newer Platform and Scribe source updates without standalone release metadata are listed first and intentionally unnumbered. Release numbers with no external documentation impact are omitted.

Current Integration Boundaries

The notes below clarify how to read this history against the public contract checked on September 6, 2026. This is a documentation checkpoint, not a new numbered product release.

Managed channel provisioning and workspace bindings are separate. Amigo supplies the provisioned use-case identifier. Public Platform clients create, inspect, or remove its service binding; the former Platform use-case inventory and explicit ownership operations are retired. See Use Cases & Bindings.
Authenticated dataset intake remains available. The separate upload-link and external HMAC paths are retired. The current intake path does not perform inline malware scanning. See Customer Data Intake and Intake Upload Links.
Standalone ringless voicemail is retired. Voicemail handling during an ordinary outbound call remains a separate voice behavior. See Voicemail.
SMS controls have distinct evidence. Registration, recipient consent, the opt-in-message send record, and suppression are separate. Provider acceptance does not prove delivery or prior consent. See SMS.

Change History

Archive Index

Older entry headings on this page remain as forwarding links for existing bookmarks. Full entries are retained in bounded archives.

WhatsApp Use Cases

Provisioning scope. This entry describes managed channel provisioning. It does not expose use-case creation or inventory through the current public workspace Platform API. Complete provisioning with Amigo, then use the service-binding contract linked above.

WhatsApp phone numbers can now be bound to use cases through the existing use case endpoints. A WhatsApp use case registers a specific phone number from a WhatsApp setup for Cloud API messaging and configures an inbound-turn webhook, following the same lifecycle as SMS, email, and iMessage use cases.

What changed:

WhatsApp channel on use case endpoints. The create, get, list, update, and delete use case endpoints now accept whatsapp as a channel value. Creating a WhatsApp use case binds a phone number from a WhatsApp setup, verifies its readiness, registers it for Cloud API use, and provisions a webhook signing secret.
One use case per phone number. Each WhatsApp phone number can be bound to exactly one active use case. Attempting to bind an already-active number returns a 409 conflict.
Phone number registration and deregistration. Creating a use case registers the phone number for Cloud API messaging. Deleting the use case deregisters it. The phone number must be verified and disconnected (not already registered under another provider) before it can be bound.
Webhook secret lifecycle. WhatsApp use cases receive a signing secret at creation, support secret rotation through the existing rotate endpoint, and re-mint the secret when the webhook URL is updated.
Setup deletion requires no active use cases. A WhatsApp setup cannot be deleted while it has active use cases. Delete the use cases first.
Phone number details in responses. Use case responses include the phone number in E.164 format and the platform-assigned phone number identifier, in addition to the webhook URL and setup reference.

What you need to do:

Create use cases for WhatsApp numbers. After completing WhatsApp setup, bind individual phone numbers to use cases to register them for messaging and configure inbound-turn delivery.
Delete use cases before deleting a WhatsApp setup. The platform now enforces that all use cases under a setup are removed before the setup itself can be deleted.
Review the Channel Manager guide for the current provisioning boundary.

Platform: WhatsApp Setup Surface (August 2026)

WhatsApp Setup Surface

Provisioning scope. Setup and provider onboarding use the managed channel authorization boundary. They are not operations in the workspace Platform OpenAPI reference. The current Channel Manager guide explains the customer integration boundary.

The Platform API now supports creating, reading, and deleting WhatsApp setups through Meta's Cloud API. WhatsApp onboarding follows a two-phase activation lifecycle where the platform creates the setup record and returns a signup URL, then Meta notifies the platform when the customer completes signup.

What changed:

WhatsApp setup endpoints. Three new endpoints - create, get, and delete - manage the WhatsApp setup lifecycle. Create registers a customer's business identity and returns Meta's Hosted Embedded Signup URL. Get returns the setup state, including whether the business account has been activated. Delete tears down credentials and webhook subscriptions.
Two-phase activation. Setups are created in a pending state. Once the customer completes Meta's Hosted Embedded Signup, the platform automatically provisions credentials and activates the channel. Callers poll the get endpoint to check activation status.
Webhook integration. The platform receives and validates account update notifications from Meta. Unrecognized or duplicate events are accepted silently to prevent retry storms.
New OAuth2 scopes. Three scopes control access: whatsapp-setup:create, whatsapp-setup:read, and whatsapp-setup:write.
One setup per business portfolio. Creating a second active setup for the same Meta business portfolio returns a 409 conflict.

What you need to do:

Grant scopes to OAuth2 clients. Clients that need to manage WhatsApp setups require the new whatsapp-setup:* scopes.
Review the Channel Manager guide for provisioning and public routing boundaries.

Platform: Entity Activity Timeline Refresh (August 2026)

Entity Activity Timeline Refresh

The per-entity activity timeline is now produced by the current-generation world pipeline, completing the migration of all world-model outputs from the previous generation.

What changed:

Timeline produced by the current pipeline generation. The entity activity timeline - used by entity detail views, FHIR patient timeline endpoints, and analytics dashboards - is now computed incrementally by the current-generation world pipeline. Previously it was produced exclusively by the prior generation.
Read-time 90-day window. The timeline retains the most recent activity events per entity (up to 200) across all time. The 90-day recency filter is now applied at read time by the serving API rather than at pipeline compute time. This means the stored timeline is generation-stable and does not require wall-clock recomputation, while API consumers continue to see the same rolling 90-day window.
Continuous incremental updates. Timeline entries update continuously as new events arrive or existing events are retracted, using the same change-driven refresh pattern as other current-generation outputs.
No schema or API changes. The timeline schema, API response shape, and query interface are unchanged. Consumers do not need to re-bind or update queries.

What you need to do:

No action required. The timeline continues to be served through the same API endpoints with the same response shape. The underlying pipeline change is transparent to consumers.

Platform: Guardian and Caregiver Entity Projection (August 2026)

Guardian and Caregiver Entity Projection

The world model now projects guardian and caregiver person entities with the same fidelity as patients and practitioners. Previously, these entities could resolve with missing name, contact, identifier, and resource type fields, which prevented downstream readers from distinguishing guardians from patients.

What changed:

Guardian entities fully projected. Person entities sourced from guardian and caregiver records now carry name, phone, email, source-system identifier, and resource type. Previously these fields could be empty, causing the entity to be indistinguishable from a patient in platform reads and agent tools.
Role discrimination corrected. Platform services and agent tools that filter person entities by role now correctly identify guardians and caregivers instead of misclassifying them as patients.
No schema changes. Guardian projections reuse existing entity columns. The stable entity table schema is unchanged, and no downstream consumer re-binding is required.

What you need to do:

No action required. Guardian and caregiver entities are projected automatically. Platform reads and agent tools that query person entities will return correct role information without configuration changes.

Platform: Generation-Agnostic Entity Cutover Support (August 2026)

Generation-Agnostic Entity Cutover Support

The stable entity publication surface now supports atomic generation cutovers. The platform can move the stable entity table between internal entity generations without downtime, consumer re-binding, or data gaps.

What changed:

Atomic generation cutover. The platform validates schema compatibility, composite key integrity, change-data-feed enablement, and publisher identity before executing a single atomic merge to converge the stable entity table with a new generation source. A postflight parity check confirms exact row-level agreement before the cutover is considered complete.
Publication is generation-agnostic. Any current or future entity generation can own publication through the same shared path. Publication is enabled per-pipeline by configuration rather than being hardcoded to a specific generation.
Metric store sources updated. Analytics models that depend on entity data now reference the stable entity table rather than generation-specific tables, ensuring metric freshness tracking remains consistent across cutovers.
No per-reader repointing. The stable publication seam remains the only supported read surface for application code. Generation-specific tables are retained for internal parity validation and rollback only.

What you need to do:

No action required. If your integrations already reference the stable entity table, they will continue to work across generation cutovers without changes. If you have saved queries that reference a generation-specific table by name, update them to reference the stable entity table advertised by the schema introspection endpoint.

Platform: Entity Reads Consolidated to Stable Publication Surface (August 2026)

Entity Reads Consolidated to Stable Publication Surface

All entity read paths - including SQL tools, analytics dashboards, research queries, and monitoring - now read from the world model's stable physical entity table. Previously, some read paths used generation-specific intermediate tables that required separate synchronization and lag monitoring.

What changed:

Single entity read surface. Platform services that query entities now bind to the stable publication table directly. This removes the previous dependency on a separately synchronized copy and eliminates an entire class of synchronization lag between entity writes and reads.
No per-reader repointing. The stable publication seam is the only supported read surface for application code. Generation-specific tables are retained for internal parity validation and rollback only - they are not advertised to consumers.
Simplified monitoring. Entity freshness monitoring now measures end-to-end latency from event ingestion to the stable entity table in a single step, rather than tracking a multi-stage pipeline across intermediate synchronized copies.
SQL tool table references updated. The fallback SQL query tool and research query schema now advertise the stable entity table. Queries that previously referenced the intermediate synchronized table should use the new table reference.
Dashboard queries updated. Built-in dashboard templates for slot analysis and facility lookup now query the stable entity table directly.

What you need to do:

Update any saved SQL queries. If you have saved queries or integrations that reference the previous entity serving table by its fully qualified name, update them to reference the current entity table advertised by the schema introspection endpoint.
No action required for standard platform usage. Dashboard templates, research queries, and tool descriptions are updated automatically.

Platform: Metric Store Execution Plan Validation and Semantic-Layer Definition Contract (August 2026)

Metric Store Execution Plan Validation and Semantic-Layer Definition Contract

The Metric Store now validates execution plans before computation and introduces a structured semantic-layer definition contract for authoring metrics as YAML.

What changed:

Execution plan validation at compute time. Before a metric computation runs, the platform validates the execution plan against the semantic-layer contract. Plans that reference an unapproved model, use an unsupported aggregation, include disallowed dimensions, or violate scoping rules (such as a global metric exposing a workspace dimension) are rejected with structured validation errors before any data is read.
Semantic-layer YAML contract. Metric definitions can now be authored as structured YAML documents that declare the metric's identity, base model, grain, dimensions, time configuration, freshness target, ownership, lifecycle status, and security classification. The platform parses, validates, and deterministically compiles these definitions into execution plans.
Supported metric types. The definition contract supports count, distinct count, sum, average, percentile, ratio, rate, score, retention, boolean, and coverage metric types. Ratio and rate metrics require separate numerator and denominator blocks.
Lifecycle statuses. Definitions carry a status of draft, certified, or deprecated.
Freshness targets. Freshness can be a named cadence (30 minutes, hourly, daily, weekly) or a positive integer of minutes.
Content-addressable definitions. Each compiled definition includes a content hash so the platform can detect changes without full re-parsing.
Scoping enforcement. Workspace-scoped metrics must include a workspace dimension. Global metrics must not include a workspace dimension. Dimensions must include a time-grain dimension.

What you need to do:

No action required for existing metrics. Metrics with valid execution plans continue to compute as before. Invalid plans that previously would have failed at query time are now rejected earlier with clearer error messages.
Use the YAML contract for new definitions. Teams authoring new metrics can use the semantic-layer YAML format for structured validation before deployment.

Platform: Metric Store Registry Migration and Global Metrics (August 2026)

Metric Store Registry Migration and Global Metrics

The metric definition registry has moved to the platform's transactional database, metric assignments now control which workspaces receive computed values, and metrics can be defined at global scope.

What changed:

Registry moved to transactional storage. Metric definitions, definition versions, and assignments are now managed in the platform's transactional store rather than in the analytical data layer. This separation gives the registry full transactional guarantees while analytical tables remain optimized for computation and querying.
Assignment-driven computation. Metric computation is now driven by explicit assignments. Each assignment links a metric definition to either a specific workspace or to global scope, and only active assignments produce computed values. This replaces the previous model where definitions were implicitly scoped to a single workspace.
Global metrics. Metrics can now be assigned at global scope. A global metric aggregates across all data without filtering to a single workspace. Global and workspace-scoped assignments for the same metric definition can coexist.
Published status supported. Metric definitions with a "published" status are now eligible for computation alongside "certified" definitions.
Legacy registry tables removed. The previous definition and version tables in the analytical layer have been added to the legacy cleanup process alongside the previously removed canonical table names.
Nullable workspace on analytical tables. The workspace identifier on metric values, freshness status, and computation run records is now optional to support global metrics that are not scoped to any single workspace.

What you need to do:

Review metric assignments. Existing workspace-scoped metrics continue to compute as before, but the assignment model is now explicit. Verify that active assignments exist for each metric you expect to see computed.
No action required for standard metric consumers. Queries against metric values continue to work. Rows for global metrics will have a null workspace identifier.

Platform: Metric Store Freshness Scheduling and Schema Cleanup (August 2026)

Metric Store Freshness Scheduling and Schema Cleanup

Metric computation is now freshness-driven with failure backoff, and superseded legacy table names have been removed from the metric store schema.

What changed:

Freshness-aware scheduling. Each computation cycle now evaluates which metric definitions are due based on their freshness target and the time of their last successful computation. Definitions that are already fresh are skipped, reducing unnecessary work.
Failure backoff. Definitions that fail computation are retried with exponential backoff starting at two hours and capping at 24 hours. This prevents transient failures from consuming compute capacity on every cycle.
Freshness status tracking. Each metric definition now tracks its last successful computation time, freshness status, and the computation run that produced the result.
Legacy table names removed. The superseded canonical table names - "decision_tool_execution_fact," "call_intelligence_fact," "event_clinical_evidence_fact," "data_source_model_health_fact," and "metric_output" - have been dropped from the metric store schema. These were replaced by the renamed models introduced in the Semantic Metric Store Foundation release.
Registry repair removed. The previously available registry repair operation has been removed. The registry now lives in the platform's transactional store and does not require manual rebuilds.

What you need to do:

Update references to dropped table names. If you have saved queries, dashboards, or external integrations that reference the old canonical table names, update them to the current names: "tool_execution_fact," "evaluation_fact," "event_evidence_fact," "model_health_fact," and "surface_interaction_fact."
No action required for standard metric definitions. Freshness scheduling and failure backoff apply automatically to all certified metric definitions.

Platform: Semantic Metric Store Foundation (August 2026)

Semantic Metric Store Foundation

The Metric Store now includes a governed semantic layer with canonical models, versioned relationships, semantic metric definitions, and automated computation.

What changed:

Canonical model registry. The platform maintains a certified registry of canonical analytical models covering conversations, interactions, tool executions, outcomes, evaluations, topic classifications, appointment workflows, entities, event evidence, model health, and surface interactions. Each model declares its grain, key columns, timestamp columns, freshness target, and source lineage.
Governed relationships. Canonical models are connected through versioned, certified relationships that declare join keys, cardinality, and allowed join direction. Metric definitions can only use relationships that are certified in the registry.
Semantic metric definitions. Metric definitions now support a semantic wrapper that binds a metric to a specific canonical model, declares approved dimensions (workspace and reporting period are required), specifies the time field and aggregation, and tracks ownership, security classification, version, and lifecycle status (draft, certified, deprecated).
Deterministic execution plans. Semantic definitions are compiled into deterministic execution plans with content hashes. The compilation validates that the base model and required relationships exist in the certified registry.
Automated computation. Certified metric definitions are computed periodically against their canonical models within a configurable time window. Results are merged using an upsert strategy keyed by workspace, metric, version, period, and dimension hash. Each computation run records its status, timing, and any error.
Freshness classification. Each canonical model and metric definition declares a freshness target. Materialized results are classified as fresh, stale, never computed, or failed based on the target and the most recent computation status.
Model renames. Several canonical models have been renamed for clarity: decision and tool execution evidence is now "tool execution fact," call intelligence evaluations are now "evaluation fact," clinical evidence records are now "event evidence fact," and data source health is now "model health fact." A new "surface interaction fact" model replaces the previous metric output model.
Metric definition versioning. Metric definitions now support independent version tracking with per-version status, content hash, execution plan, and lifecycle timestamps.

What you need to do:

No action required for existing integrations. The semantic layer is additive. Existing metric definitions and values are unaffected.
Review canonical model names. If your integration references canonical model names in metric definitions, update to the new names: use "tool_execution_fact" instead of "decision_tool_execution_fact," "evaluation_fact" instead of "call_intelligence_fact," "event_evidence_fact" instead of "event_clinical_evidence_fact," and "model_health_fact" instead of "data_source_model_health_fact."

Platform: Design Agent Collector Endpoint Removed (August 2026)

Design Agent Collector Endpoint Removed

The design-agent session collector endpoint has been removed from the Platform API.

What changed:

POST /v1/design-agent/session removed. The write-only endpoint that accepted design-agent session transcripts is no longer available. Requests to this path now return 404.
Collector authentication scheme removed. The dedicated bearer-token authentication scheme used exclusively by the design-agent collector is no longer part of the API. This token was separate from platform API keys and granted no read access.
No replacement endpoint. This endpoint served an internal collection workflow and has been retired without a public replacement.

What you need to do:

Remove any integrations that call the design-agent session endpoint. If you have scripts or hooks that POST to /v1/design-agent/session, remove or disable them. The endpoint no longer exists and calls will receive a 404 response.
Discard any stored collector tokens. The dedicated write-only token for this endpoint is no longer valid for any purpose.

Platform: Date Range Filtering on List Sessions Endpoint (August 2026)

Date Range Filtering on List Sessions Endpoint

The list sessions endpoint now supports optional date range filtering so callers can retrieve sessions created within a specific time window.

What changed:

created_after query parameter added. An optional ISO 8601 UTC timestamp that acts as an inclusive lower bound on session creation time. When provided, only sessions created at or after this timestamp are returned.
created_before query parameter added. An optional ISO 8601 UTC timestamp that acts as an exclusive upper bound on session creation time. When provided, only sessions created before this timestamp are returned.
Half-open range convention. The two parameters form a half-open interval [created_after, created_before), consistent with the convention used by the appointments endpoint. Either parameter can be supplied independently.
Validation. When both parameters are supplied, created_after must be strictly before created_before. If this constraint is violated, the endpoint returns a 400 Bad Request with the code invalid_date_range.

What you need to do:

No action required for existing integrations. The parameters are optional and the default behavior is unchanged. Add created_after and/or created_before query parameters to narrow results by creation date when needed.

Platform: RelatedPerson Resource Type and Patient Filtering (August 2026)

RelatedPerson Resource Type and Patient Filtering

The FHIR surface now supports the RelatedPerson resource type for guardians and contacts, and Patient queries no longer return non-patient persons.

What changed:

RelatedPerson added as a supported FHIR resource type. Guardians and contacts are now projected as RelatedPerson resources with their own identity, name, telecom, and relationship data. Previously, these persons had no dedicated FHIR representation.
Patient results no longer include guardians. Patient searches now exclude persons whose role is guardian or contact. Previously, guardians without clinician-specific attributes were returned as patients, producing patient records with no birth date or medical record number.
Resource identity uses raw vendor identifiers. FHIR resource ids for patients, practitioners, and related persons now use the vendor's native identifier directly, without a namespace prefix. This ensures that external identity resolution matches the identifiers presented during session creation. Existing resources will resolve under the new identifiers.
Summary counts reflect three-way role split. The resource summary endpoint now counts Patient, Practitioner, and RelatedPerson as separate categories, matching the filtering applied to search results.

What you need to do:

Update integrations that consume Patient bundles. If your integration expected guardian records in Patient results, query RelatedPerson instead.
Update stored resource references. If your integration stores FHIR resource ids and previously expected a namespace prefix on patient, practitioner, or guardian ids, update to expect the raw vendor identifier.
No action required for standard FHIR queries. Searches against Patient, Practitioner, and RelatedPerson return correctly scoped results without additional filtering.

Platform: Workspace Environment Field Removed from API (August 2026)

Workspace Environment Field Removed from API

The environment field has been removed from workspace API requests and responses. Workspaces no longer carry an environment designation at the API level.

What changed:

environment removed from workspace creation. The create-workspace request body no longer accepts an environment field. Previously, callers could pass production, staging, or development as an environment designation; the value was accepted but had no operational effect. The field is no longer recognized.
environment removed from workspace update. The update-workspace request body no longer accepts an environment field. Previously, the field was accepted and silently ignored. It is no longer recognized.
environment removed from workspace responses. Workspace response objects no longer include an environment field. Previously, the response reported the serving deployment's stage as the workspace environment. This field is no longer returned.
Environment conversion endpoints removed. The GET /{workspace_id}/environment-check and POST /{workspace_id}/convert-environment endpoints have been removed. These endpoints previously provided a pre-check and a no-op conversion between staging and production environments. They are no longer available.

What you need to do:

Remove environment from request bodies. If your integration sends an environment field when creating or updating a workspace, remove it. The field is no longer accepted.
Update response parsing. If your integration reads the environment field from workspace responses, remove that dependency. The field is no longer returned.
Remove environment conversion calls. If your integration calls the environment-check or convert-environment endpoints, remove those calls. The endpoints no longer exist and will return errors.

Platform: Inline Malware Scanning Removed from Intake Upload Path (August 2026)

Inline Malware Scanning Removed from Intake Upload Path

The inline malware scanning step has been removed from the intake file upload path. Uploads are no longer scanned synchronously before persistence.

What changed:

Inline scan removed. The upload endpoint no longer performs a synchronous malware scan before writing the file. Previously, when scanning was enabled, every upload was scanned inline and rejected if the scanner returned a positive detection or was unavailable (fail-closed). This step has been removed entirely.
Scanner-related error responses removed. The upload endpoint no longer returns 422 Unprocessable Entity for malware detections or 503 Service Unavailable for scanner unavailability. These responses are no longer possible.
Scanner configuration removed. The environment variables that controlled inline scanning enablement, host, port, and timeout are no longer recognized. Deployments that set these variables can remove them with no effect.
No change to upload validation. File size limits, SHA-256 deduplication, and CSV validation continue to operate as before.

What you need to do:

No action required for most integrations. If your error-handling code checks for 422 responses with a malware-scan detail message or 503 responses for scanner unavailability on the upload endpoint, those branches are now unreachable and can be removed.
Remove scanner environment variables. If your deployment sets intake scanner configuration variables, they can be safely removed.

Earlier entry links (preserved bookmarks)

Metric Store Setup, Refresh, and Run Status Endpoints

Continue with the archived entry.

Entity Overlay Available on Text and Web Session Open

Continue with the archived entry.

Cost-to-Serve Endpoints for LLM Inference and Infrastructure

Continue with the archived entry.

Workspace Region and Environment Are Now Deployment-Derived

Continue with the archived entry.

Canonical Metric Store Models

Continue with the archived entry.

Workspace Data Query Tools Available in Trigger Actions

Continue with the archived entry.

Ringless Voicemail Channel Removed

Continue with the archived entry.

Cross-Entity Projection Parity and Retraction Correctness

Continue with the archived entry.

Remaining World Model Readers Migrated to Current Serving Projection

Continue with the archived entry.

World Model Read Path Migrated to Current Serving Projection

Continue with the archived entry.

World Model Stateful Projections Upgraded

Continue with the archived entry.

Legacy Per-Workspace Drive Credential Path Removed

Continue with the archived entry.

Auto-Bind Identity Resolution Matches on Data Source Name

Continue with the archived entry.

Guardians Emitted as First-Class Person Entities

Continue with the archived entry.

Design Agent Collector Gating Moved to Client

Continue with the archived entry.

Connector Source Types Consolidated

Continue with the archived entry.

Production Evaluations Limited to Metric Definitions

Continue with the archived entry.

Read-Side Observability for Enrichment Overlay

Continue with the archived entry.

Synchronous Enrichment Overlay on Entity Read

Continue with the archived entry.

Structured Metric Verdicts in Evaluations

Continue with the archived entry.

Intake Storage Simplified to Per-Tenant Volumes

Continue with the archived entry.

Unused Data Connectors Removed

Continue with the archived entry.

Canonical ROS-Grid Cell Shape for Structured Notes

Continue with the archived entry.

Raw Note-Generation Output Archived for Debug and Audit

Continue with the archived entry.

Automatic External Identity Binding on Person Fold

Continue with the archived entry.

Shareable Upload Links and Launch DPC Job Removed

Continue with the archived entry.

World Model Serving Layer Upgrade

Continue with the archived entry.

Entity Enrichment Read Path Repoint

Continue with the archived entry.

World-Model Projection Efficiency Improvements

Continue with the archived entry.

Entity Overlay Write Path Wired to Enrichment PUT

Continue with the archived entry.

Synchronous Entity Enrichment Overlay

Continue with the archived entry.

Result-Retaining Interactive Write Idempotency

Continue with the archived entry.

Unified Enrichment Winner Selection

Continue with the archived entry.

Ephemeral Tool Results Honored on Text and Simulation Sessions

Continue with the archived entry.

Improved Circuit Breaker Resilience for Voice Calls

Continue with the archived entry.

Improved Session Liveness Detection for In-Person Recordings

Continue with the archived entry.

Info Section Coding Echo Removed from AMD Note Templates

Continue with the archived entry.

Psych-Progress and Therapy-Progress Emergency/Telehealth Field Updates

Continue with the archived entry.

Psychiatry Intake Vitals Grid and BP Field Updates

Continue with the archived entry.

Psychiatry Intake ROS Header and Option Labels

Continue with the archived entry.

Therapy Note Template Updates - Generated Fields, Section Renaming, and Version Bumps

Continue with the archived entry.

Psychiatry Progress Note Restructuring - Field Moves, Editable Vitals, and Field Removal

Continue with the archived entry.

Psychiatry Progress Note Field Relocation and Version Bump

Continue with the archived entry.

Psychiatry Progress Note Section Restructure and Version Bump

Continue with the archived entry.

AMD Template Parity - Multi-Answer Conditions, Exam Grid Update, Field Renaming, and Version Bump

Continue with the archived entry.

AMD MBC Parity - Editable Scores, Treatment Plan Checkboxes, and Version Bump

Continue with the archived entry.

Ephemeral Tool-Result Persistence

Continue with the archived entry.

Production Evaluations On by Default

Continue with the archived entry.

HMAC External Upload Path Removed

Continue with the archived entry.

Patient-Only Warehouse Mart Support

Continue with the archived entry.

Data Warehouse Connector Source Type

Continue with the archived entry.

Keyless Credential for Google Drive Intake Sources

Continue with the archived entry.

Second Diagnosis-Confirm Gate for Therapy Progress Documentation

Continue with the archived entry.

CPT Location Gates, Charge Coding, and Interventions Grid for Documentation Templates

Continue with the archived entry.

End-to-End Session Traceability in Observability Logs

Continue with the archived entry.

Voice Turn Authentication Fix for Multi-Workspace Deployments

Continue with the archived entry.

Runtime Toggle for Finalize Content Validation

Continue with the archived entry.

Stronger Model for Note Generation

Continue with the archived entry.

Per-Section Confirmation No Longer Required Before Finalize

Continue with the archived entry.

AI Metric History Now Accumulates Across Runs

Continue with the archived entry.

Distinct Error When Zoom Meeting Has Not Started

Continue with the archived entry.

Design Agent Session Collector

Continue with the archived entry.

Per-Source AI Call Attribution for Cost Caps

Continue with the archived entry.

Topic Modeling Gated by Permission

Continue with the archived entry.

Provider-Authored ICD-10 Codes and Generation-Decoupled Code Reads

Continue with the archived entry.

Circuit Breaker on World Events Serving Reads

Continue with the archived entry.

Operator Dashboard Timeline Reads Migrated to Optimized Serving Path

Continue with the archived entry.

Remaining Timeline Readers Migrated to Optimized Serving Path

Continue with the archived entry.

Analytics Summary Reads Migrated to Optimized Serving Path

Continue with the archived entry.

Per-Workspace Timeline Read Migration to Materialized View

Continue with the archived entry.

No-Diagnosis Attestation Waives Diagnosis Finalize Gate

Continue with the archived entry.

Session Codes Status Reflects Generation Outcome

Continue with the archived entry.

PUT Allowed for Structured-Note Autosave

Continue with the archived entry.

MRN Prefix Removed from Seeded Appointment Charts

Continue with the archived entry.

Seed Appointments Use Eastern Time Wall-Clock Scheduling

Continue with the archived entry.

Structured Finalization Gates

Continue with the archived entry.

Zoom Session Finalization for Silent Meetings

Continue with the archived entry.

Improved Transcript Retrieval Reliability

Continue with the archived entry.

Mid-Session Auto-Check No-Op Guard

Continue with the archived entry.

Server-Authoritative Checklist Item Category

Continue with the archived entry.

Faster Mid-Session Transcript Snapshots

Continue with the archived entry.

Structured Grounding for Dependent Generations and Section Regeneration

Continue with the archived entry.

Schema-Constrained Structured Note Generation

Continue with the archived entry.

Typed Structured-Note Envelope and Body Deprecation

Continue with the archived entry.

All-Session AMD Template Pinning on Create and Update

Continue with the archived entry.

Authoritative AMD Note Template Registry

Continue with the archived entry.

Cancelled Sessions Hidden from Session and Appointment Lists

Continue with the archived entry.

Patient Chart Fields on Appointments

Continue with the archived entry.

Automatic Retry for Telephony Provisioning Status Callbacks

Continue with the archived entry.

Server-Side Checklist Seeding at Session Creation

Continue with the archived entry.

Resilient Transcript Retrieval Under Degraded Storage Conditions

Continue with the archived entry.

Improved Observability for Transcript Retrieval Failures

Continue with the archived entry.

Empty Extraction Results Treated as Valid

Continue with the archived entry.

Improved Diagnostic Detail for Structured Generation Failures

Continue with the archived entry.

Robust JSON Parsing for Structured Generation Stages

Continue with the archived entry.

Session-Owned Note Generation Fields

Continue with the archived entry.

Zoom Meeting Bot Shutdown Drain Reliability

Continue with the archived entry.

Zoom End-of-Session Transcript Completeness and Graceful Degradation

Continue with the archived entry.

Zoom OAuth Scopes Updated for RTMS Event Delivery

Continue with the archived entry.

Zoom Bot Request Parity with BFF

Continue with the archived entry.

Generic Session Create Scoped to In-Person Only

Continue with the archived entry.

Zoom OAuth Token Auto-Refresh and Reconnect Detection

Continue with the archived entry.

Automatic Note Generation for In-Person Mic Sessions

Continue with the archived entry.

Transcript Retrieval Resilience Improvement

Continue with the archived entry.

Voice Turn Accept Header Content Negotiation

Continue with the archived entry.

Assist Features - Action Items, Ask, Auto-Check, and Section Regeneration

Continue with the archived entry.

Simulation Step Timeout Increased

Continue with the archived entry.

Workspace Provision Endpoint Reschedules Scope Provisioner

Continue with the archived entry.

World Model Workspace-Scoped Entity Projection

Continue with the archived entry.

Scheduled Workspace Topic Classification

Continue with the archived entry.

World Model Pipeline Event Log

Continue with the archived entry.

Authorization-Bound Test Values in Simulations

Continue with the archived entry.

Post-Visit Mutations, Note Versioning, and Finalize Gates

Continue with the archived entry.

Async Generation Orchestration

Continue with the archived entry.

Zoom Session Live Event Stream

Continue with the archived entry.

Zoom Session Lifecycle and Bot Controls

Continue with the archived entry.

Zoom OAuth Granular Scopes

Continue with the archived entry.

Zoom Integration Connect and Disconnect

Continue with the archived entry.

Attach-Ticket Endpoint and CORS Hardening

Continue with the archived entry.

SQL Authorization Binding Test Values for Platform Functions and Workspace Data Queries

Continue with the archived entry.

Run-Scoped Token Caching for Integration Calls

Continue with the archived entry.

Scribe Sessions Now Born In-Progress

Continue with the archived entry.

Emotion Detection Now Requires GPU and Runs in Half-Precision

Continue with the archived entry.

Intake Source Secret Namespace Change

Continue with the archived entry.

Mesh-Only Certificate Authentication for Internal Service Tokens

Continue with the archived entry.

Mutual TLS Client Authentication for Service Credentials

Continue with the archived entry.

Async Audit Exports and Export Catalog Removal

Continue with the archived entry.

Cross-Workspace Audit Log Reads

Continue with the archived entry.

Enterprise SSO, SAML/OIDC Federation, and MFA Retired

Continue with the archived entry.

Per-Workspace IP Allowlist Removed

Continue with the archived entry.

Per-Workspace Conversation Topic Analytics Controls

Continue with the archived entry.

Per-Turn Authorization for Durable Text Sessions

Continue with the archived entry.

Authorization-Scoped Parameter Execution for SQL Tools

Continue with the archived entry.

Access Token Lifetime Restored to 15 Minutes

Continue with the archived entry.

Async Audit Export with Presigned Download Links

Continue with the archived entry.

Per-Workspace Intake Deduplication Override

Continue with the archived entry.

Source Provenance on Batch Document Intake

Continue with the archived entry.

FHIR Patient Bundle Builder Endpoint

Continue with the archived entry.

Default Access-Token Lifetime Extended to 24 Hours

Continue with the archived entry.

Authorization-Bound Parameter Schema Hiding

Continue with the archived entry.

Authorization-Bound Parameters for Platform Functions and Workspace Data Queries

Continue with the archived entry.

Extended Default Token TTL for Email OTP Provider Grants

Continue with the archived entry.

Receipt-Backed Background Polling and Delivery Acknowledgement

Continue with the archived entry.

Generate ICD-10-CM Code Suggestions

Continue with the archived entry.

Zoom Active-Session Guard Narrowed to Live-Capture States

Continue with the archived entry.

Zoom Session Lifecycle and Transcript Finalization

Continue with the archived entry.

Connector Sync-Run Telemetry Routing Change

Continue with the archived entry.

Appointments Endpoint Returns a Rolling Date Window

Continue with the archived entry.

Provider Grants No Longer Require a Pre-Existing Provider Entity

Continue with the archived entry.

Session Lifecycle Writes, Mode, and Idempotency

Continue with the archived entry.

Nested Session Object on Appointments

Continue with the archived entry.

Appointments Endpoint - Current-Day Listing and Retrieval

Continue with the archived entry.

Admin Provider Access Grant Management for Scribe

Continue with the archived entry.

Source Provenance and PHI-Gated Filenames in the Analytical Catalog

Continue with the archived entry.

Provider-M2M Act-As-by-Email Delegation

Continue with the archived entry.

External Auth Claims Mapping for Customer-Attested Authorization

Continue with the archived entry.

Shared LLM Client Lifecycle for Production Evals and Insights

Continue with the archived entry.

Approval Rejection Closes Approval State Cleanly

Continue with the archived entry.

Voice Provider Prompt Aligned with Runtime Tool Capabilities

Continue with the archived entry.

Streaming Attach-Ticket Handshake Enforcement

Continue with the archived entry.

Token Exchange - Browser Attach Tickets for Streaming Sessions

Continue with the archived entry.

Provider M2M Client Credentials - Machine-to-Machine Provider Tokens

Continue with the archived entry.

Internal Auth Simplification - Shared Secrets Retired for Cluster-Internal Dispatch

Continue with the archived entry.

Environment Variable Cleanup - Four Runtime Knobs Retired

Continue with the archived entry.

Empty Tool Description Warning for Realtime Voice Sessions

Continue with the archived entry.

Graph Tool Instructions Preserved for Realtime Voice Sessions

Continue with the archived entry.

Structural Tool-Calling Enforcement for Realtime Voice Sessions - Reverted

Continue with the archived entry.

Autonomous Tool Calling for Realtime Voice Sessions

Continue with the archived entry.

Per-Modality Voice COGS Tracking

Continue with the archived entry.

Workspace-Level Allocation Rate Limit and Fleet Exhaustion Visibility

Continue with the archived entry.

External Identity Binding Documentation Consolidated

Continue with the archived entry.

Environment Configuration Simplified - Volume Paths, Transcript Settings, and Connector URL

Continue with the archived entry.

Drive Sync Resilience - Interrupted Syncs No Longer Strand Files

Continue with the archived entry.

Simplified Rolling-Transcript Storage Key Layout

Continue with the archived entry.

Feature Gates Retired - Six Capabilities Now Always On

Continue with the archived entry.

Required Tool Retry for Realtime Voice Sessions

Continue with the archived entry.

Voice Tool Authorization Aligned Across Playground and Phone Calls

Continue with the archived entry.

Playground Voice Test Calls Use the Configured Voice Provider

Continue with the archived entry.

Exhaustive Skill Reference Scanning on Delete

Continue with the archived entry.

Tiered Permission Enforcement for Workspace Data Query Invocation

Continue with the archived entry.

Permission-Based Access Control for API Keys and MCP Tools

Continue with the archived entry.

Compliance Routes Require Audit.view Permission

Continue with the archived entry.

External Identity Binding for Memory v2

Continue with the archived entry.

Session Allocation for Streaming Workers

Continue with the archived entry.

Resumable WebSocket Streaming for Browser Recording Sessions

Continue with the archived entry.

Use-Case Ownership Endpoints and List Proxy Removed

Continue with the archived entry.

Unified Default and Custom Memory Dimensions

Continue with the archived entry.

Patient-Scoped Memory Expansion Tool

Continue with the archived entry.

Version-List Endpoints Scoped to Caller's Workspace

Continue with the archived entry.

This page covers v0.9.400 and later. Older releases live in the archive pages: v0.9.250 - v0.9.399, v0.9.100 - v0.9.249, and v0.6.0 - v0.9.99.

Path-Bearing MCP Resource URIs and SAML SP Base Decoupling

Continue with the archived entry.

Healthie Inbound Connector

Continue with the archived entry.

Version-List Endpoints Scoped to Caller's Workspace

Continue with the archived entry.

Canonical Pagination for List Endpoints

Continue with the archived entry.

OAuth Authorization Server Discovery and Updated Model Routing

Continue with the archived entry.

External Identity Bindings for Returning Users

Continue with the archived entry.

Scribe Notes, Summaries, and Checklists

Continue with the archived entry.

Event-Based Triggers Now Match Live Workspace Events

Continue with the archived entry.

Conversation Starters and Typed Real-Time Voice Controls

Continue with the archived entry.

Evaluation Quality Trends and Conversation-Level Simulation Verdicts

Continue with the archived entry.

Durable Dataset Updates and Cloud-Storage File Export

Continue with the archived entry.

Classic Conversation History Import

Continue with the archived entry.

Unified Run Detail, Trajectory, and Non-Voice Authored Turns

Continue with the archived entry.

API-Key Permission Catalog

Continue with the archived entry.

Per-Question (Turn-Level) Metering Attribution

Continue with the archived entry.

Universal Metering - All Workspaces Metered

Continue with the archived entry.

Interactive API Documentation (ReDoc)

Continue with the archived entry.

Google Drive Intake - Native File Export and Shortcut Resolution

Continue with the archived entry.

Optional Domain-Wide Delegation for Google Drive Intake Sources

Continue with the archived entry.

AI-Powered Clinical Note Generation and Finalization

Continue with the archived entry.

Configurable Output Voice for Real-Time Speech-to-Speech

Continue with the archived entry.

Default Realtime Voice Model Updated

Continue with the archived entry.

Per-Conversation Token Attribution for Text Channels

Continue with the archived entry.

Bulk Import Endpoint for Conversation Migration

Continue with the archived entry.

Channel-Agnostic Eval Quality Analytics Endpoint

Continue with the archived entry.

Durable Background Reply Delivery, SMS Opt-In, Turn Idempotency, and Conversation Close

Continue with the archived entry.

Analytics Dashboard KPI Endpoint Typed and Operator Performance Aggregation Fixed

Continue with the archived entry.

Legacy Read Endpoints Removed Behind Unified Runs Surface

Continue with the archived entry.

Per-Call Media Isolation Now Universal

Continue with the archived entry.

API-Key Role and Permission Catalog Endpoint

Continue with the archived entry.

Framework Run Trajectory Endpoint

Continue with the archived entry.

Intake Dataset Update Runs - One-Click Sync, Prepare, and Publish

Continue with the archived entry.

EHR Appointment Reschedule Reliability - Patient Identity and Edit Verification

Continue with the archived entry.

Durable EHR Appointment Sync, Delete Detection, and Faster Detail Refresh

Continue with the archived entry.

EHR Patient Creation No Longer Auto-Assigns Intake Packet Cohort

Continue with the archived entry.

Streaming Intake File Downloads

Continue with the archived entry.

Single-Run Detail Endpoint

Continue with the archived entry.

Non-Voice Takeover Route Dispatch and Authored-Turn Endpoint

Continue with the archived entry.

Non-Voice Operator Takeover - Authored-Turn Send-Gate

Continue with the archived entry.

Multi-Value OR Filters on Runs List and Summary

Continue with the archived entry.

Scribe Sessions API

Continue with the archived entry.

Paused Conversation Run Status

Continue with the archived entry.

Run-Scoped Switch-Mode and Access-Token

Continue with the archived entry.

Simulation Recovery and Metering Accuracy

Continue with the archived entry.

Run-Scoped Takeover and Handback

Continue with the archived entry.

Promote Playground Session to Coverage Run

Continue with the archived entry.

Live Built-in Tool Catalog

Continue with the archived entry.

Run-Scoped Operator Guidance

Continue with the archived entry.

Unknown sort_by Fields Now Rejected with 422

Continue with the archived entry.

Skill Sampling Parameters Threaded Through Tester

Continue with the archived entry.

Simulation Run Failure Reasons and Suite Completeness

Continue with the archived entry.

Drain-Aware Simulation Executor with Instant Re-Drive Handoff

Continue with the archived entry.

Simulation Run Re-Drive and Automatic Recovery

Continue with the archived entry.

Orphaned Simulation Run Reclaim and Graceful Cancellation

Continue with the archived entry.

Dynamic Behaviors Feature Removed

Continue with the archived entry.

Billable Token Metering on Raw Real-Time Voice Runtime

Continue with the archived entry.

Reasoning Effort Pinned for Realtime Reasoning Models

Continue with the archived entry.

Full-Duplex Voice Family Is Generally Available

Continue with the archived entry.

Guidance Delivery Status Honesty, Transcript Read-Audit, and Blueprint GA Status

Continue with the archived entry.

Durable Audit Event for Operator Whisper Guidance

Continue with the archived entry.

Surface Responses Expose Review and Archive Provenance

Continue with the archived entry.

Audit Export Downloads, Actor Email on Audit Entries, Permission Enforcement

Continue with the archived entry.

Turn Identity on Streaming Done Frame

Continue with the archived entry.

Voice Model Family Rename

Continue with the archived entry.

Full-Duplex Voice Model Family

Continue with the archived entry.

Stable Turn Identity on Conversation Turns

Continue with the archived entry.

Text Conversation Lifecycle Hardening

Continue with the archived entry.

Normalized Checks View for Simulation Evaluation Results

Continue with the archived entry.

Escalation Policy and Risk Signal Configuration Removed from Service API

Continue with the archived entry.

Per-Turn Prompt Log Emission for Voice Calls

Continue with the archived entry.

LLM Token Metering for Framework Agent Runs

Continue with the archived entry.

Audio Verification Removed - Reduced Per-Turn Voice Latency

Continue with the archived entry.

Returning-User Entry State for Known Callers

Continue with the archived entry.

Legacy Clinical Event Review Surface Retired

Continue with the archived entry.

External Write Proposal Review API

Continue with the archived entry.

External Write Proposal Delivery

Continue with the archived entry.

Self-Serve Custom Memory Dimensions via Enrichment Key Tags

Continue with the archived entry.

Real-Time Live Voice Overlay on Runs List and Summary

Continue with the archived entry.

Unified Runs Summary Endpoint

Continue with the archived entry.

Simulation Runs Summary Endpoint

Continue with the archived entry.

Per-Request Wait-for-Final and Filler Suppression for Background Tools

Continue with the archived entry.

JWT Bearer Integration Auth - RFC 7523 Flow Selection

Continue with the archived entry.

Server-Side Operator Identity Enforcement

Continue with the archived entry.

Event-Driven Delivery for Background Tools on Non-Live Channels

Continue with the archived entry.

Identity Binding Test Values in Simulation Sessions

Continue with the archived entry.

Eager Post-Conversation Evaluation for All Channels

Continue with the archived entry.

Triggerable Event Type Validation

Continue with the archived entry.

Trigger Run Observability Metrics

Continue with the archived entry.

Durable Trigger Run Pipeline

Continue with the archived entry.

SMS Pre-Send Content-Compliance Check

Continue with the archived entry.

Eager Post-Call Production Evaluation

Continue with the archived entry.

Run List Enrichment - Entity Name and Service Name

Continue with the archived entry.

Trigger Run Idempotency Guard

Continue with the archived entry.

Durable Trigger Run Processing

Continue with the archived entry.

Run List Enrichment - Caller, Direction, Turns, and Outcome

Continue with the archived entry.

Unified Runs List Endpoint

Continue with the archived entry.

Terminal Lifecycle Marker for Failed and Timed-Out Framework Runs

Continue with the archived entry.

Tenant Isolation for Session Event Injection

Continue with the archived entry.

Provider Access Grant API with Role-Based Scopes

Continue with the archived entry.

Scribe Provider Scopes

Continue with the archived entry.

Managed Integrations

Continue with the archived entry.

Agent Runs - Durable List Endpoint

Continue with the archived entry.

Voice Conversations - Per-Turn Transcript Recovery

Continue with the archived entry.

Agent Runs - Harness Context Endpoint

Continue with the archived entry.

Agent Runs and Definitions - Drop CrewAI Framework Support

Continue with the archived entry.

Agent Runs - Durable Trajectory Persistence

Continue with the archived entry.

Agent Runs - Native Definition Runs End-to-End

Continue with the archived entry.

Agent Definitions - Native Agent Definition Registry

Continue with the archived entry.

Customer Data Intake - Folder Path Materialization

Continue with the archived entry.

Memory - Clinical State Projection (Phase-2b Structured Dimensions)

Continue with the archived entry.

Metering - Daily Meter Grain, Invoice Summation Fix, and Connector Enrichment Metering

Continue with the archived entry.

Memory - Custom (Layer-2) Dimensions in Extraction

Continue with the archived entry.

Agent Runs - Native Definition Validation

Continue with the archived entry.

Agent Runs - Native Run Mode

Continue with the archived entry.

Agent Runs - Actor Attribution and Multi-Agent Trajectory

Continue with the archived entry.

Intake Files - Hide Withdrawn Documents

Continue with the archived entry.

Agent Runs - Dual-Mode Auth, Cache Usage, and Hardening

Continue with the archived entry.

OIDC/SAML Federation - Consumed-Invitation Recovery

Continue with the archived entry.

Workspace-Bound Admin Routes and platform:admin Scope

Continue with the archived entry.

Simulation Performance - Per-Run Metric Breakdown

Continue with the archived entry.

Simulation Performance Overview - Aggregated Run Analytics

Continue with the archived entry.

Agent Runs - Framework Agent Execution

Continue with the archived entry.

Fleet Status - Tool-Runner Fleet Observability

Continue with the archived entry.

Cross-Provider Identity Linking for Invited Operators

Continue with the archived entry.

Source Change Routing - Document Withdrawal and Cursor Persistence

Continue with the archived entry.

Folder-Listing Diff Deletion Fallback for Folder-Only Access

Continue with the archived entry.

Incremental Change Detection for Drive-Based Data Sources

Continue with the archived entry.

Epic SMART Backend Services JWKS Endpoint

Continue with the archived entry.

Trigger Run History - Durable Execution Log for Triggers

Continue with the archived entry.

Document Embeddings in Intake Materializer

Continue with the archived entry.

Intake Materializations List Endpoint

Continue with the archived entry.

Use Case Ownership Endpoints

Continue with the archived entry.

Intake Materialization Endpoint

Continue with the archived entry.

Async Tool Lifecycle Signaling

Continue with the archived entry.

Current Integration Boundaries

Change History

Archive Index

WhatsApp Use Cases

WhatsApp Setup Surface

Entity Activity Timeline Refresh

Guardian and Caregiver Entity Projection

Generation-Agnostic Entity Cutover Support

Entity Reads Consolidated to Stable Publication Surface

Metric Store Execution Plan Validation and Semantic-Layer Definition Contract

Metric Store Registry Migration and Global Metrics

Metric Store Freshness Scheduling and Schema Cleanup

Semantic Metric Store Foundation

Design Agent Collector Endpoint Removed

Date Range Filtering on List Sessions Endpoint

RelatedPerson Resource Type and Patient Filtering

Workspace Environment Field Removed from API

Inline Malware Scanning Removed from Intake Upload Path

Metric Store Setup, Refresh, and Run Status Endpoints

Entity Overlay Available on Text and Web Session Open

Cost-to-Serve Endpoints for LLM Inference and Infrastructure

Workspace Region and Environment Are Now Deployment-Derived

Canonical Metric Store Models

Workspace Data Query Tools Available in Trigger Actions

Ringless Voicemail Channel Removed

Cross-Entity Projection Parity and Retraction Correctness

Remaining World Model Readers Migrated to Current Serving Projection

World Model Read Path Migrated to Current Serving Projection

World Model Stateful Projections Upgraded

Legacy Per-Workspace Drive Credential Path Removed

Auto-Bind Identity Resolution Matches on Data Source Name

Guardians Emitted as First-Class Person Entities

Design Agent Collector Gating Moved to Client

Connector Source Types Consolidated

Production Evaluations Limited to Metric Definitions

Read-Side Observability for Enrichment Overlay

Synchronous Enrichment Overlay on Entity Read

Structured Metric Verdicts in Evaluations

Intake Storage Simplified to Per-Tenant Volumes

Unused Data Connectors Removed

Canonical ROS-Grid Cell Shape for Structured Notes

Raw Note-Generation Output Archived for Debug and Audit

Automatic External Identity Binding on Person Fold

Shareable Upload Links and Launch DPC Job Removed

World Model Serving Layer Upgrade

Entity Enrichment Read Path Repoint

World-Model Projection Efficiency Improvements

Entity Overlay Write Path Wired to Enrichment PUT

Synchronous Entity Enrichment Overlay

Result-Retaining Interactive Write Idempotency

Unified Enrichment Winner Selection

Ephemeral Tool Results Honored on Text and Simulation Sessions

Improved Circuit Breaker Resilience for Voice Calls

Improved Session Liveness Detection for In-Person Recordings

Info Section Coding Echo Removed from AMD Note Templates

Psych-Progress and Therapy-Progress Emergency/Telehealth Field Updates

Psychiatry Intake Vitals Grid and BP Field Updates

Psychiatry Intake ROS Header and Option Labels

Therapy Note Template Updates - Generated Fields, Section Renaming, and Version Bumps

Psychiatry Progress Note Restructuring - Field Moves, Editable Vitals, and Field Removal

Psychiatry Progress Note Field Relocation and Version Bump

Psychiatry Progress Note Section Restructure and Version Bump

AMD Template Parity - Multi-Answer Conditions, Exam Grid Update, Field Renaming, and Version Bump

AMD MBC Parity - Editable Scores, Treatment Plan Checkboxes, and Version Bump

Ephemeral Tool-Result Persistence

Production Evaluations On by Default

HMAC External Upload Path Removed

Patient-Only Warehouse Mart Support

Data Warehouse Connector Source Type

Keyless Credential for Google Drive Intake Sources

Second Diagnosis-Confirm Gate for Therapy Progress Documentation

CPT Location Gates, Charge Coding, and Interventions Grid for Documentation Templates

End-to-End Session Traceability in Observability Logs

Voice Turn Authentication Fix for Multi-Workspace Deployments

Runtime Toggle for Finalize Content Validation

Stronger Model for Note Generation

Per-Section Confirmation No Longer Required Before Finalize

Informed Consent Registered as a Standalone AMD Section

AI Metric History Now Accumulates Across Runs

Distinct Error When Zoom Meeting Has Not Started