See the What's New section for an overview of what was released last month.

This week's release notes include:

  • Copilot
  • AI Agents and Knowledge
  • Messaging and live chat
  • App Marketplace
  • Zendesk QA
  • Products with no updates this week

Copilot

New:

  • If auto assist uses similar tickets to generate a suggestion, then these similar tickets will show up in rationale under sources.
  • Agent edits of auto assist's reply suggestions, as well as removals of actions suggested by auto assist, now show up in the ticket event log
  • On the Intent page, it's now possible to see key metrics about how many tickets were automated with intents, how many intent conflicts you have, and how many new intent suggestions you have for your account.
  • Increased the number of custom intents that are possible to create to 500.

AI Agents and Knowledge

New:

  • Knowledge Blocks now support two outcomes: “Answer found” and “No answer found.” This allows you to define what should happen when no relevant answer is retrieved (for example, escalate, ask a follow-up question, or route to a fallback step). Existing procedures have been automatically migrated and continue to work as before—this enhancement simply adds more flexibility.

  • The Knowledge Agent is now controlled through a new Knowledge Procedure, giving you full control over how knowledge-based responses are handled. You can now:

    • Add follow-up questions (“Did that answer your question?”)
    • Branch the conversation based on user responses
    • Trigger additional actions or flows

    By default, if no answer is found, the agent will fall back to your default reply (same behavior as before).

  • Customers who want to run the Knowledge Builder again can go directly to the URL '/knowledge/generation/wizard' and generate more knowledge.
  • The Knowledge Editor now allows up to 2,000 excluded terms instead of 300.

Fixed:

  • Voice – Knowledge Agent access from procedures: Fixed an issue where the Knowledge Agent, when triggered from a procedure, could incorrectly respond that it could not access the knowledge base.
  • Voice – Escalation audio overlap/race condition: Resolved edge cases where escalation audio could overlap with ongoing conversational responses, causing double or conflicting messages.

Messaging and live chat

New:

  • The custom queues API is now live, enabling customers to retrieve queue metrics (Inbound and outbound traffic count, acceptance rate, queue wait time, and queue size) using the API for email, messaging, and talk channels.

App Marketplace

New:
  • Flint AI Assistant (Support)
    • Flint AI Assistant is a native Zendesk Support app that gives agents an AI-powered copilot inside the ticket sidebar. It reads your help center articles, uploaded PDFs, and connected knowledge sources, then drafts accurate replies grounded in your documentation — not generic web results. Ask Flint a question from any ticket. It retrieves the most relevant articles using semantic search (RAG) and synthesizes a citable answer in seconds. Responses stream in real time.
  • Convert to Internal by ZENCLAB (Support)
    • Convert to Internal by ZENCLAB is a simple, straightforward app that helps you save time. Stop manually converting public comments to internal notes — let the app do it for you. Set up a task in seconds: choose how often it runs, add filters if you like, and you're done. The app works quietly in the background while you focus on what matters. Or hit Run Now whenever you need an instant conversion.
  • AI Customer Simulator (Support)
    • AI Customer Simulator turns your Zendesk training tickets into live practice scenarios. When an agent replies to a tagged training ticket, the AI responds as the customer — contextually, realistically, and with natural escalation patterns. Agents get real conversation practice. You get measurable training outcomes. No real customers involved.
  • AI Ticket Generator (Support)
    • AI  Ticket Generator reads your published Help Center articles and uses AI to generate believable inbound support tickets — complete with varied customer personas, writing styles, intents, and urgency levels. No more empty sandboxes. No more manually writing fake tickets. Just realistic test data that reflects your actual product and customers.
  • Graphio.ai (Support)
    • Graphio.ai is a metadata-only analytics platform that predicts broken handoffs and cross-team misalignment before they cost you revenue.  Graphio.ai analyzes API-enabled metadata from your existing systems to automatically identify misalignments among Sales, CS, Legal, Finance, Product, and Operations before deals stall, renewals fail, or contracts slip. Technically, it’s a Context-Change Graph AI that learns your company's blueprint for success. It tracks live performance against that map to predict, alert on, and repair dangerous deviations the moment they occur.
  • PilotFlow Lite (Support)
    • PilotFlow Lite is a lightweight automation app designed to extend Zendesk's native capabilities with more flexible and practical rule-based workflows. With PilotFlow Lite, Zendesk admins can automatically assign tickets, apply tags, and add followers based on conditions that are not supported by default triggers. This includes rules based on requester email domains, number of open tickets per requester, ticket activity patterns, and more.
  • Staircase AI (Support)
    • Staircase AI integrates with Zendesk to fetch support ticket information, including ticket volume, subjects, and comments. Once connected, Staircase AI pulls all new tickets created by your clients into the Activity tab, allowing you to see how many new tickets are opened each week and on what topics. This completes your view of account communications by adding your support channel alongside other relationship data, providing analysis such as sentiment, statistics, and insights while eliminating manual work.

Zendesk QA

New:

  • Removed console warnings from app navigation for a cleaner console and better debugging experience.
  • Disabled the evaluate button when reviews are from the shared pool. This prevents users from evaluating reviews not assigned to them.
  • Hid dispute actions from the tasks view
  • Added evaluations to activity notifications. Users now get notified about evaluation activities.
  • Implemented count-based fetching for ticket retrieval, resulting in better performance when loading tickets.
  • Allow score-only evaluations with an empty comment object. Users can now submit evaluations without comments when appropriate. 

Fixed:

  • QA no longer persists events for duplicate survey answers. This helps prevent data duplication.
  • To help maintain accurate timestamps, QA now preserves answered_at when CSAT custom field overrides satisfaction rating.
  • Return AccountUser and MonitorWorkspace for user detail and workspace API for more complete API responses for user and workspace data. 

Products with no updates this week

  • Admin Center
  • Support
  • Voice
  • AI agents - Advanced
  • Analytics
  • Zendesk Workforce Management (WFM)
Powered by Zendesk