Release notes

Updates to the MCP Server documentation and tools:

get_improvement_guide returns Ada’s entity improvement guide — when to use articles, coaching, playbooks, and other entities, plus best practices for making suggestions. Call it before using propose_change.

send_feedback lets AI assistants send structured feedback directly to Ada’s development team — for workflow gaps, tool quality issues, or missing capabilities.

propose_change now uses a three-phase flow: discover field schemas, preview the proposed change for user confirmation, then execute with confirmed: true.

Learn more in the MCP Server documentation.

The v1.conversation.handoff.ended webhook now carries the Conversations API custom channel ID, extending channel filtering to handoff-ended events.

What’s new

This change brings handoff-ended events in line with the other Conversations API webhook events that already support channel filtering.

• Channel ID on handoff-ended events. When a handoff ends on a Conversations API custom channel, the v1.conversation.handoff.ended event is now tagged with both the custom channel’s ID and the handoff integration name. Previously it carried only the handoff integration name.
• Filter handoff traffic by channel. An endpoint scoped to a custom channel ID now receives that channel’s handoff.ended events. Conversations API handoff traffic can be isolated by channel even when other channels hand off to the same integration.

Native Chat, Voice, and Email handoff-ended events are unchanged.

For details, see Webhook channel filtering.

The Webhooks dashboard now displays the live Sunshine Conversations webhook configuration for AI Agents with a SunCo channel connected, including the target URL, subscribed triggers, and whether the configuration uses Multi-bot Routing.

This makes it possible to verify a SunCo integration’s routing pattern — for example, when adding a second AI Agent to an existing SunCo app — without inspecting the SunCo API or production database.

The card appears under Config > Platform > Webhooks for AI Agents with a SunCo channel; AI Agents without one see no change.

For more information, see Sunshine Conversations setup.

API key names can now be edited from the dashboard. Previously a name was fixed at creation, and correcting a typo or repurposing a key meant rotating to a new key and updating every integration.

See Edit an API key name.

Voice now supports 42 of Ada’s 60 supported languages, up from 8.

Newly supported Voice languages

The following 34 languages are now available for Voice conversations:

Albanian, Arabic, Bosnian, Bulgarian, Catalan, Chinese (Simplified), Chinese (Traditional), Croatian, Czech, Danish, Estonian, Filipino (Tagalog), Finnish, Greek, Hebrew, Hindi, Hungarian, Indonesian, Japanese, Korean, Lithuanian, Macedonian, Malay, Norwegian, Polish, Romanian, Russian, Serbian, Slovak, Slovenian, Thai, Turkish, Ukrainian, Vietnamese.

These join the 8 languages already supported on Voice: Dutch, English, French, German, Italian, Portuguese, Spanish, Swedish.

Configuration

Configure newly supported languages on the Ada dashboard under Config > CHANNELS > Voice:

• Assign a default language to each phone number, SIP domain, or Twilio Application.
• Add languages to the language selection menu so callers can pick their preferred language at the start of a call.
• Select a speaking voice for each enabled language on the Speaking voice tab.

For the full Voice-supported language list and channel-by-channel coverage, see About multilingual support.

The maximum size of profile.metadata on the End Users API has been increased from 4 KB to 10 KB.

This applies to both:

• POST /v2/end-users/
• PATCH /v2/end-users/{end_user_id}

The cap is now enforced consistently on both endpoints. Requests with metadata over 10 KB return 400 with "Metadata exceeds the maximum size of 10 KB." No changes are required on your end; existing integrations sending smaller payloads are unaffected.

For full details, see the End Users API reference.

The Data Export API now includes a new message type, tool_call, that records when an AI Agent selects a tool during a conversation. It currently covers Actions and Handoffs configured on the AI Agent.

See the message object reference for the schema.

The Conversations endpoint in the Data Export API (v2) now exposes a new used_coaching field.

The field returns a list of Coaching events that were applied during the conversation. Each entry includes the coaching event id, entity_name, coaching_type, coaching_instructions, and coaching_intent. Conversations without Coaching return an empty array.

For more details, see the conversation object reference.

Coaching is now supported on simulated conversations so you can proactively adjust AI Agent behavior.

Coaching applied to a simulated conversation will be available for the AI Agent to use in production conversations. All coaching types are supported: send a message, handoff, use knowledge, actions, process, and run a playbook.

Learn more in the Simulations documentation.

The Messages endpoint in the Data Export API (v2) now exposes additional message data:

• A new sms_consent_asked message type, indicating that an end user was asked to consent to receiving SMS messages.
• A new sms_text field on the csat_shown message, returned when the CSAT survey was delivered via SMS. The field contains the exact text that was sent to the end user.

For more details, see the message object reference.