We have added a new optional field to the Email Conversations API:
cc: A comma-separated list of email addresses to be added as CC recipients to the Email conversation.This field allows you to specify additional recipients that should be CC’d on the AI Agent’s email response.
The Data Export API now returns autocapture variables within the variables attribute of each conversation object. Previously, this attribute contained only global variables.
This enhancement is available across all versions of the API and requires no configuration changes.
We have added two new optional fields to the End User profile object:
avatar: A URL to the end user’s avatar imagedisplay_name: The end user’s display nameWe’ve added new events to the subscribeEvent action in Embed2 to help you better track user interactions within the chat window:
If you’re using the Embed2 script, these events are immediately available. For the NPM package, update to the latest version (1.9.0) to start using them.
The language field on the article object in the upsert API now supports BCP 47 language tags (e.g., en-CA, fr-FR). The previously supported locale field is deprecated but still accepted for backwards compatibility if language is not provided.
API responses from GET /v2/knowledge/articles and GET /v2/knowledge/articles/{id} will continue to return locale, but it is no longer documented. In these responses, the language field contains the ISO 639-1 two-character language code, not the full BCP 47 tag.
This is part of a gradual migration. Eventually, locale will be removed from both persisted records and API responses, and language will consistently use BCP 47 tags across all endpoints.
We’ve added a new event to the subscribeEvent action in Embed2. With this event, you can receive a callback when an end user submits a CSAT. The event returns a numeric CSAT score:
If you’re using the Embed2 script, these events are immediately available. For the NPM package, update it to the latest version (1.8.16) to start using them.
Validation added to reply_to field in Email Conversation API
Validation for the reply_to field in the Email Conversation API will be introduced in the coming weeks. Once implemented, requests with an invalid email address will fail with a validation error.
Ada is releasing a new version (v2) of our platform APIs. For detailed instructions on how to migrate your integrations, please see our Migration Guide.
In addition to an updated developer documentation site, we’re adding the following capabilities to our platform in this release:
API Endpoints Consolidation
Several legacy endpoints have been merged and restructured by resource and action. This makes integrations simpler and reduces redundancy. For example:
https://example.ada.support/api/end-users/v1/https://example.ada.support/api/v2/end-users/Streamlined API Tokens
A single API token now works across all endpoints. Previously, separate tokens were required per API. This change simplifies token management and authentication.
Uniform Response Structures
Responses (including error messages) are now standardized. This reduces the complexity of handling different response formats and improves overall consistency.
Standardized Pagination
Pagination parameters and response metadata are now consistent, enhancing the developer experience when handling large data sets.
Improved Rate & Data Limits
More transparent rate and data limit policies have been introduced. This ensures better reliability, prevents unexpected throttling, and makes it easier to plan request volumes.
Data Export API Updates
The Data Export API endpoints have been updated to follow the new v2 naming convention:
/api/data_api/v1.4/conversations → /api/v2/export/conversations/api/data_api/v1.4/messages → /api/v2/export/messagesBackward-Incompatible Changes
Some v1 functionality has been deprecated or altered. Review your current integrations and ensure you’re using the appropriate v2 endpoints and parameters.
We’ve released some security improvements to Embed2.
Previously the setSensitiveMetaFields action in Embed2 action and sensitiveMetaFields setting were only supported in scripted bots, now it is supported in Generative AI Agents.