
Zendesk News Roundup for October 2025
Zendesk October updates: unified navigation across apps, ISO 42001 AI certification, AI Agents expand to email/webforms, intelligent voice triage, dynamic lookup filters, record-level permissions, and improved brand and user management for smoother, more consistent workflows.
Hey. Welcome to the October roundup newsletter.
Next week Zendesk is presenting its AI Summit, their big fall event in which they'll announce new features and launch new capabilities to their Resolution Platform.

Last year brought us auto-assist for Agent Copilot, the new Zendesk Voice and the zero training AI Agent Advanced. If you're interested, follow the link below to register for the digital broadcast, or keep an eye out for a full article on the releases later this week!
Save your agents valuable time looking for duplicate users and keep your user data clean, unified and consistent.
Our app automatically detects and displays potential duplicate users right in the ticket sidebar. Agents can then merge them with a single click, keeping your customer data clean and consistent.
Zendesk
Unified navigation

Zendesk is pleased to announce a modernized, unified global navigation model across Zendesk products. We’re updating the product interfaces to provide a consistent, simplified approach to navigation.
The Zendesk platform exists out of a collection of different tools. Some are build in house, others were acquisitions that got a partial Garden redesign and a few UI elements that made it fit with the family.
To name just a few: Agent Workspace, Admin Center, Knowledge, Analytics (Explore), AI Agents Advanced (Ultimate), QA (Klaus), WFM (Tymeshift). Each of these has its own UI and logic, with the Zendesk navigation elements added to the top or side of their existing UI,
Later this year Zendesk will update all their applications to have a unified navigation across all tools. The header will get a consistent location for user profiles and application switching, and the secondary sidebar navigation will get the same experience across all apps with a collapsable state, same color and icon treatment and uniform search placement.
I hope the new UI will still allow for adding a touch of personal brand color though. All screenshots I've seen have the same grey colors, and I kinda like the brand color to differentiate between environments.
ISO42001 Certification
AI is transforming how businesses engage with customers, but adoption hinges on trust. IT and security leaders know that without clear safeguards, AI can introduce new risks – ranging from data privacy concerns to opaque decision-making. That’s why we’re proud to announce that Zendesk is one of the first CX companies to achieve ISO 42001 certification, the world’s first international standard for AI management systems.
In short this means enabling Zendesk AI in your company will be a lot easier from a compliancy and security standpoint. The entire article on Zendesk's blog is worth a read, I learned a lot about this new standard by doing so!
AI Agents & Messaging
AI Agents for webforms and email
Zendesk used to be a self service solution that used Help Center articles as a way to provide customers answers to their questions. Across the Help Center (obviously..), email, web forms and chat, customers got a list of links to relevant content after submitting their question.
With the advent of AI, replying to those inquiries has shifted away from links to articles towards custom generated replies to each unique customer question based on the knowledge stored in your Help Center. First available for Messaging, and later for the Help Center via Quick Answer, this capability has now also been made available for email replies.
Every Zendesk Suite customer now has access to generative responses across all included channels, giving them a taste of how AI can help them automate ticket resolutions.

Read my full overview in last weeks' article
AI Agent Advanced updates
Zendesk's AI Agent Advanced got a series of small quality of life updates this month.
For one, there's a new dashboard that greats you when you login that shows you an overview of your AI Agents across all channels and some basic metrics. There's also a new analytics dashboard that gives you insight into the most important high level metrics.



Additionally, the new Agentic Procedures now have versioning, giving you more insight and control over changes made to your flows. If changes lead to less effective conversations you can easily revert to a better version, or see what changes were made over time and how they did improve your resolution numbers.
Agent Workspace & Routing
Agent Timeout
Omnichannel Routing routes tickets to available agents with the right skillset and with available capacity. That availability depends on an agents' state. And while agents can put themselves online, offline or online via a dropdown menu in Agent Workspace easily, it's easy to forget doing so when going away for lunch or a short break.
Idle timeout has long been a way to automatically change an agents status when they're gone from their device. Zendesk detects this state by a lack of keyboard or mouse input. There's however a second reason an agent might disappear and that's when their internet connection drops, or if they turn off their computer.

With the updates Status timeout settings for Omnichannel routing you can now choose to set a different status based on an agent going idle or disconnecting. You can for example set them to away if they go idle and se them to offline if they disconnect. Depending on your Omnichannel routing settings, the idle state will retain their assigned tickets, while the offline state will reassign their tickets to other available agents if a customer replies.
Closing tickets automation
Zendesk is changing how the default automation Close ticket 4 days after status is set to solved behaves. Instead of only acting on tickets at the top of the hour, this automation will now close tickets throughout the day as it finds tickets that are eligible based on the automation configuration. Tickets will continue to be closed after the period defined in each account’s solved-to-closed automation settings.
I wonder if this is done for efficiency reasons to spread the database load, for reporting reasons (to spread closure times across the hour instead of peaks every hour) or a prelude to giving us more fine-grained automation rules that allow for automations based on minutes or specific times of day. I assume the former, but I really hope it's the latter.
Internal Note trigger action
Previously, internal notes added by ticket triggers were attributed to the ticket assignee at the time the trigger fired. This release changes that so they are more accurately attributed to the system user in the ticket events and reporting.
System is indeed a more honest way of representing these actions. Instead of inheriting the name of the last agent looking at the ticket, this automations are now attributed to the system.
Similarly, with the advent of AI Tickets in Agent Workspace, actions done by the AI Agent will be tagged as such too, giving you more auditable insight on who did what. Would be nice if we could attribute actions to named apps and integrations too to get even more honest insights.
Side Conversation updates
Side Conversations got a few quality of life updates this month:
Only show agent email addresses, which allows admins to restrict which addresses are autocompleted when adding a user to a side conversation. This setting is off by default.
This will prevent accidentally sharing customer interactions with third parties with similar names to your teams. Would be nice if you could also allow specific organizations (suppliers, consultants, external HR and Finance teams ...) too.
Allow dynamic subdomain detection, which lets admins loosen certain requirements on how their Teams tenants are set up. This setting is recommended only when standard troubleshooting can’t resolve an intractable problem, and is off by default.
No idea what this means, but if you're using Teams instead Slack, I hope this helps 😅
Use signatures in side conversations, which lets admins to control whether signatures are included in side conversations, and whether agent or brand signatures are used. This setting is on by default.
A big finally. Plenty of customers asking for this!

Agent Copilot
Intelligent Triage for Voice

We’re excited to introduce intelligent triage for Voice calls. This helps you get structured insights on intent, entity, sentiment, and language that power smarter post-call automations, richer reporting, and continuous workflow improvements.
Pretty cool upgrade to Zendesk Voice. After a conversation wraps up Zendesk used the call transcript to detect the Intent and Sentiment of the customer during the call. Useful for further routing if followup via email is required, or a good way to get a more complete report across all channels.
Knowledge & Insights
Service Catalog
The Service Catalog for Employee Service is now generally available to all customers using Zendesk Suite or higher. New in this release is the ability to add images and rich text to service descriptions, as well as icons to differentiate the different services in the catalog itself.

If you want to learn more about the new Service Catalog and Approvals, take a look a this preview I published after Relate 2025.

Platform & Security
Brand settings
With the introduction of Departments and the new segmented access to tickets based on brands, Zendesk has undoubtedly breathed new life into the old brands feature.
And while the ability to split ticket access by brands is by far the biggest change, the upcoming End User Separation is another big change that will further segment Zendesk data across brands in the near future.
This month saw a couple of other nice additions to brands management. For one, you can now choose if customers see only tickets for a specific brand, or tickets across brands in the Help Center request portal. For traditional brand setups this might not be useful, but if you have an IT and HR department (brand) in your instance, you might want your employees to be able to see their support requests across all teams in one place, regardless of the Help Center their looking at.

From a management perspective you can now choose to assign new agents to all brands automatically, or keep the manual option where you can select which brands each agent should have access too.

And to make management even easier, there's now a new brands option on Agent profiles to update their access directly from their support profile in Agent Workspace.
Editing user permissions
You can now give more granular rights to agents when managing user profiles. You can choose between read only, editing, or full rights that allow for deletion too.
Previously this setting was linked with editing organizations and groups, but now these permissions are split, giving you more specific configuration options to manage custom roles.

Dynamic filtering for Lookup Fields
Lookup Fields are a powerful ticket field type that allow you to link objects in Zendesk together. You can have an Account Manager Lookup Field on the Organization level that lists users tagged with "Account Manager".
Or you can have a lookup field "Product" that allows you to link a Custom Object record to a ticket.
Until now these Lookup Fields could be filtered based on users (owner is the current requester) or organizations (SLA contract assigned to this organization).

With the new dynamic filters you can filter Lookup Field based on any value. If you have a large database of IT assets, you can create a lookup field that only lists devices of type laptop.

Similarly you can now link different Lookup Fields. If you first create a field that lists Product Types, you can then create a Lookup Field that lists all devices that match that type.
Ticket settings

With the new page layout, related settings are grouped together so they can be quickly found at a glance. Setting sections can be expanded or collapsed as needed to reduce page length and visual overload.
These new settings are a lot more structured and easier to grasp for new Admins. I like it!
Looking at the Admin Center overall, it feels a lot more structured now, although I still get lost between Messaging and Omnichannel Routing settings. And Automations and Dynamic Content are still an eyesore to look at.
Expiring Attachments

Speaking of the new ticket setting layout, look at how clear this new Downloads and attachments section looks. And the updated design is not the only new element here. We can now add an expiration date to attachments, removing the ability to download them after a set amount of hours.
Custom Object updates
Lookup Fields with Dynamic relationships mentioned earlier are especially powerful in combination with Custom Object records. However these are not the only changes to Zendesk's Custom Objects
Incremental Export for Custom Objects

It is now possible to incrementally export records from any custom object using new APIs designed for external reporting and data pipelines. The export returns all of the specified custom object's records that were created, updated, or deleted since the defined start time.
For customers with large sets of data, or data that changes a lot, the old paginated API that allowed for export per 100 elements wasn't ideal. This new endpoint allows you to export all changes from a specific point in time, making it way easier to keep other platforms in sync with Zendesk, or build an event stream that acts upon certain changes.
GET /api/v2/incremental/custom_objects/{custom_object_key}/cursor?start_time={start_time}
Record level permissions

Until now, permissions have been defined at the object level, meaning each role either had access to all of an object's records or none of them. Now, admins can define granular access rules for each custom role's ability to create, read, update, and delete custom object records. Rules are defined for each action individually, so a single custom role could have different access rules applied to their permissions to create an object's records and update an object's record.
Record-level permissions make it possible for admins to protect sensitive information in custom objects and ensure it is surfaced to only the agents and end users who need to access it.
Similar to the aforementioned Lookup Field filters this is yet another very powerful capability useful for some scenarios. This access rule gives specific user roles access to only a specific set of objects. You might have a support team that one needs access to Active customers. Or your Facility team in Antwerp only needs to see service requests that are for that location.
These new access rules allow you to give specific access to users based on the data in records, making sure your teams only see what they need to see.

Major Changes
Moving AI agent access into Zendesk Roles
Ultimate, and by extension AI Agents Advanced had its own user and permissions model that gave your team access to your AI Agents to manage it.
Starting in October Zendesk will move those users into Zendesk and your team can only access AI Agents Advanced with a proper Zendesk account similar to how you would give them access to Analytics or the Help Center.
This does not mean however you need to buy additional seats for these agents. Apparently you can combine a Light Agent or Contributor role in Zendesk with Admin permissions for AI Agents.

End of life for Zendesk Sell
Zendesk Sell will be retired on August 31, 2027. While this means that we’ll no longer provide a sales CRM solution, we’re committed to ensuring Sell customers are well supported during this period.
To help ensure a smooth transition to another solution, (...), we’ve partnered with Pipedrive (...)

I liked Sell as a basic CRM but I'm glad Zendesk's focus shifted back to CX and EX.
End of life for custom resolutions in AI agents

Custom resolution states are being removed from AI agents - Advanced. We will also be removing the Custom resolutions tab in the Analytics dashboard.
Automated resolutions are a better measure of your organization’s automation success than custom resolutions. Automated resolutions evaluate whether customer queries truly get addressed by your AI agents.
I always like the small difference between informed and resolved when building custom flows. Informing the customer that "sadly you can't return an item after more than 30 days" is not the same as resolving their "this HDMI adapter I bought last month doesn't work with my new Nintendo Switch" question.
That being said, having the system automatically categorize escalated vs resolved vs unresolved will make it a bit faster to write procedures or build dialogues.

Perspectives
Average handling time

Yes, metrics like average handle time and after-call work still matter for human-powered interactions. But at an aggregate level, success is increasingly about the speed and quality of resolutions – whether driven by automation, agents, or a combination of both.
If AI Agents can respond immediately and can scale so it doesn’t matter if a customer talks for 2, 5 or 10 minutes with them, metrics like average handling time and fist reply time don’t matter anymore
And consequently, if AI handles the easy part, what will actually happen is that, when looking at your human team in isolation, those values actually go up. Those escalated conversations that do require a human touch often require more time to resolve.
So you need new metrics. Like quality of the conversation. How the customer feels after their issue is resolved. Effectiveness and problem solving.
AI Databases

It’s also agents creating agents, and agents creating applications, and agents deciding they need a database to complete a task. The explosion of these smart machine uses and workflows is well underway. But we don’t have an infrastructure that was made for that world. They were all designed to work with humans.
One of the main weaknesses of current AI Agents for customer support solutions is that they are limited to indexing and understanding knowledge. They’re awesome at converting your existing documentation into generative replies to customers.
But some questions require other types of data to resolve them. Order status, product comparisons, error log analysis. Zendesk procedures and actions are a step towards getting that data actionable, but there’s still big steps to make here.
Speaking of other types of data, take a look at this new Agentic Commerce initiative from Stripe and OpenAI.

Top retail automation hacks
Great overview of how you can use integrations to improve your retail support by my friends from Knots and AgnoStack leveraging integrations with your web shop, using AI to read and detect order numbers in communication and tons of other great Zendesk tips.

Empathic AI
Under Tom’s leadership, Zendesk has embraced a simple mantra: The customer is always human. He’s quick to admit the future is changing fast: Within five years, “over 80% of customer interactions” could be handled by AI agents, and in half of cases, customers will bring their own AI agents too.
That doesn’t mean the human element disappears. Quite the opposite. In his view, AI should make things faster, more personal, and easier to predict, while still leaving room for people to choose how they engage. - Source
And finally...
Did you know Zendesk's AI Agent is called Zea?
