Welcome to Ada’s release notes. Scroll down to see a list of recent releases.

You can subscribe to updates with our RSS feed, or sign up to get news about updates directly in your inbox.

At the end of every week that has had at least one feature release, we’ll send you an email on Friday at 11 a.m. Eastern to let you know about our last few releases.

Chat settings update

Updated the Chat settings area, reorganizing settings for: launching Ada, managing chat persistence, and controlling privacy and security.

Previous organization:

  • Launch controls
  • Persistence
  • Privacy & security
    • Privacy: IP tracking
    • Configuration: Enable chat
    • Approved domains

New organization:

  • Launch
    • Chat availability: Enable chat
    • Allowed websites (formerly “Approved domains”)
    • Launch controls
  • Data & privacy
    • Persistence (formerly its own “Persistence” tab)
    • Privacy: IP tracking
    • Approved link protocols

Custom link protocols

Configure custom link protocols so that your AI Agent can send deeplinks, such as yourbrand://your-app.

By default, your AI Agent can already send links with these standard protocols: https, http, ftp, ftps, mailto, tel, callto, sms, cid, xmpp.

What’s new

Add custom link protocols (URI schemes) for your AI Agent to use in your chat settings.

Why this matters

This allows your AI Agent to send deeplinks so that your customers can more easily take action to address their issue.

Support for Multiple Participants in Email Conversations

Your AI Agent now seamlessly supports email conversations involving multiple participants!

When additional recipients are included — whether in the “To” or “CC” fields — the AI Agent keeps the exchange in a single thread and can respond to inquiries from any participant. This ensures clearer, more coherent conversations for everyone involved.

Learn more about Email conversations with multiple participants here.

Introducing Playbooks

Enterprises invest heavily in customer service, but complex inquiries still rely on human agents, because traditional bots can’t follow real-world standard operating procedures (SOPs).

Playbooks empowers your AI agent to follow your standard operating procedures, handling complex, multi-step inquiries with the same precision and flexibility as your best human agents. That means:

  • Interpreting instructions in real time
  • Adapting to unique customer inquiries
  • Using Actions to get information and perform tasks
  • Referencing Variables for personalization
  • Escalating when needed

Like your SOPs, Playbooks are written in natural language with step-by-step instructions, making them easy for you to review and maintain over time. We’ve also made it easy for you to create your first Playbooks through a simple prompt or an existing PDF.

Learn more about Playbooks in our launch blog post, documentation, and best practices guide.

Limitations

While Playbooks offer powerful capabilities, it’s important to be aware of their current limitations:

  • Not supported on Ada Voice: Playbooks currently only work with Messaging and Email channels.
  • Limited Coaching support: While you can coach the AI Agent on when and which Playbook to use, Coaching is not considered once the AI Agent is executing a Playbook. Additionally, you cannot currently coach individual steps or messages taken during a Playbook. To adjust the AI Agent’s behaviour in these cases, edit the Playbook directly.
  • Limited knowledge access: The AI Agent currently cannot initiate a search for knowledge articles while using a Playbook. If relevant knowledge would be helpful, consider incorporating it directly into the Playbook content.

Being aware of these limitations helps ensure realistic expectations and supports better use of Playbooks in your workflows.

New report: Proactive conversations

We’ve launched a new Proactive conversations report to help you better understand how your AI Agent is engaging customers through Proactive messaging—and how those conversations are performing.

This report gives you:

  • A line graph comparing your total conversation volume with the number of Proactive conversations (those that customers responded to).
  • A table showing key metrics for each Proactive message.

With this new report, you can quickly identify which Proactive messages are driving engagement, resolution, and satisfaction—and which ones might need adjustment.

For more information, see this topic.

Improved SMS Consent Messaging for Voice Conversations

Your Ada AI Agent now provides clearer information when requesting permission to send text messages during voice conversations. When your AI Agent asks for SMS consent, it will specify that it wants to send a “single, one-time text message” rather than just a “text message”.

This improvement helps set clear expectations for customers about the nature of SMS communications they’ll receive, ensuring they understand they’re consenting to individual messages rather than ongoing text conversations.

Learn more about how SMS consent works in Voice conversations.

Localized Knowledge Articles

Knowledge articles are now tagged with localized BCP 47 language codes (e.g., en-US) instead of the root language name (e.g., English).

What’s new

  • BCP 47 language codes – Articles now use BCP 47-compliant codes, so you can distinguish content for specific regions.
  • Locale-aware retrieval – When users have a localized language tag, Ada automatically surfaces matching localized articles.
  • Region-specific Coaching – When Coaching the AI Agent, you can pick region-specific articles by language tag (e.g., en-US vs. en-GB).
  • Create localized articles in Ada – Author localized articles directly in Ada, tagged with the appropriate BCP 47 code.
  • Upsert localized articles via the Knowledge API – Add or update localized articles programmatically through the Knowledge API.

Limitations

  • Knowledge-base integrations remain single-locale – Integrations still sync only one locale per language. If your knowledge base has both en-CA and en-US articles, Ada will sync only one locale.
  • Limited end-user localization – Ada cannot detect users’ regions automatically. Most users will have only the root language tag (e.g., en). In these cases Ada searches articles from all locales for that language unless you add availability rules to narrow results.

Why this matters

Companies serving multiple regions that share a language can now store and deliver region-specific knowledge, ensuring every user sees content that’s accurate for their locale.