Ada release notes

We're excited to introduce several enhancements that improve how your AI Agent understands, responds, and engages customers—leading to better interactions that increase AI Agent CSAT and Automated Resolution.

What’s new:

• Smarter clarifications: AI Agents now handle ambiguous questions more effectively and provide thoughtful clarifications to guide customers through possible topics.
• Improved empathy and follow up questions: AI Agents are better at acknowledging customer issues and following up with a customer after providing solutions to offer additional help.
• More relevant and comprehensive: AI Agents are better at referencing the most relevant articles and reference full articles instead of snippets for more comprehensive answers.
• Inline Links: Links in Ada Chat will now be rendered inline to improve readability.

We're excited to introduce Coaching - the first tool in our product that can be used to teach AI Agents how to improve over time. Coaching lets you provide feedback on what your AI Agent says, thinks, and does - enabling it to learn from past behaviour and improve.

What’s different:

• Coaching: From the conversation view, you can now Coach directly on the AI Agent's reasoning decisions, searched knowledge, action and process selection, and generated responses.
• Conversation Log: The Conversation UI has been reimagined to provide more context on the reasoning steps the AI Agent undertook and to optimize this experience for providing feedback.
• Custom Instructions: We’ve repositioned Guidance as 'Custom Instructions' and have migrated any of your existing Guidance to the new Custom Instructions tab under the AI Agent Profile. Note that any older Guidance will not experience any change in the performance based on the transition to Custom Instructions, but some of your Guidance may be improved by recreating as Coaching. For more details, see our Help Doc on Recreating Contextual Custom Instructions as Coaching

For more details on best practices for Coaching and how to teach your AI Agent, check out the Help Doc on Teaching your AI Agent How to Improve with Coaching

Topics are critical for creating an effective feedback loop! As such, we’ve introduced several reporting enhancements to help you manage and interpret Topic performance more efficiently.

What’s new:​

1. Category aggregation:
   • Category-level performance metrics: Roll-ups from Topics within the category are now available directly on the main Topics report.
   • New drill-through Category pages: Just like a single Topics page, Category-specific pages aggregate insights across all Topics rolled up under that Category.
   • Category filters: Filter conversations and the AR report by Category for more detailed analysis.
2. Download the entire Topics report, now including Category data.
3. Locked table headers keep column names visible as you scroll.

We've optimized the way Ada indexes and searches for relevant knowledge when writing replies to your customers. After internal testing, we expect this will result in more accurate and relevant answers, as your AI Agent is more likely to retrieve the correct content. No action is required to take advantage of this update.

We've updated the end user experience in Chat when speaking with an AI Agent. Now, your end users will be presented with a new indicator whenever the AI Agent is generating a response. Users will also see an updated chat footer informing them that the AI Agent is powered by Ada.

These changes apply to web, iOS, Android, and React Native SDKs.

1. New Knowledge Usage Report - will help you understand which articles are most frequently used by Ada, and which articles are correlated with high or low Automated Resolution Rates.

2. New Actions Usage Report - gives you visibility into how often Ada is performing each action, and highlights errors with full log download functionality - allowing your team to troubleshoot effectively.

3. New Knowledge and Action Filters in Conversation View: New filtering capabilities allow users to drill into conversations where specific knowledge articles or actions were referenced, giving deeper insights to inform an effective feedback loop.

4. Topics Reporting Enhancements: Enables users to drill down from high-level topic performance data to see top utilized articles and actions in conversations classified under the same topic. This visibility greatly enhances the improvement loop by providing direction on what optimizations to focus on to improve performance for a given topic.

Note: Historical data will go back as far as Oct 1st, 2024.

We're excited to introduce six new languages added to Ada's multilingual translations:

• Afrikaans
• Kinyarwanda
• Macedonian
• Swahili
• Xhosa
• Zulu

You can now auto-translate or create custom translations to better serve customers who contact you over Messaging or Email in any of the above languages.

If you use a Smart Capture block to capture a number, numeric string, or phone number from a caller, your callers will now be able to provide the value by speaking it or by entering it into their phone's dial pad.

Before capturing a value, your AI Agent or bot will automatically tell callers that they can either say the number or enter it into their dial pad.

For more information, check out our Help Docs:

• If you have a generative AI Agent, see Use the Smart Capture block to intelligently capture callers' responses.
• If you have a scripted bot, see Use the Smart Capture block to intelligently capture callers' responses.

Resetting your domain is now possible on the Email channel configuration page. Delete your existing domain and generate new DNS records to give your AI Agent the ability to respond on behalf of a different domain/email address.

Learn more about how to change your email domain at Bring Your Own Domain.

We're excited to introduce a new level of flexibility in how your AI Agent captures information with Autocapture Variables.

For each autocapture variable you can select one of the following:

• Ask when needed (default): The AI Agent will prompt customers for a value only if it can’t infer it from the conversation.
• Always ask: Your AI Agent will always ask the customer to provide a value. If used in a Process, the customer will be asked each time the Process runs.

You now have the ability to select this behavior when creating an Autocapture Variable or editing an existing one, giving you more control over how your AI Agent interacts with customers.

To try it out, head to the Create Variable or Edit Variable screens, where you’ll find the new Capture Behavior configuration option.