AI

Hi Zendesk.Could you have additional filters for the Merge ticket suggestions so we can set a hard rule to only show merge suggestions if the tickets have same intent and if tickets belongs to same group. Additionally, having options to restrict this feature for certain roles would be great.Thanks

Dentro do agent IA Advanced tem resultados de relatorio que não me ajudam a ter detalhes do que preciso. Preciso utilizar estes resultados do Bot dentro do Anality Explore para ter resultados personalizados e resultados que não tenho dentro do Agent IA Advanced Resultados como uma visão de retenção, BSAT mais detalhado, mapa de calor mais no sentido de acompanhamento ter um verdadeiro: REALTIME dos temas mais falando, o que está acontecendo no dia e quando eu acessar etc… ter dashboard´s personalizados do Bot para saber o que está acontecendo e ter a visão do que realmente eu preciso ter internamente para controle e curadoria do Bot entre outras frentes. muito razo os relatorios do agent IA Advanced preciso personalizar!

Hi, currently Voice AI can only look into Zendesk KB, the Crawler are limited to 20, which is too less.Are Advanced AI Agent customer get access to this platform to have all Crawler and sources available for Voice AI?Thanks.Tobias

We have three different role types in our software, which are basically:- basic end user- manager level- admin level And each role type has a different experience in the software. We don't necessarily know which role a person is in, so AI doesn't always answer correctly.  Even as a human agent, it's tough: sometimes you can tell based on context. Sometimes you can look it up in the system. If they have written in before, we may have stored it in their profile (we have a custom field for that).  Any tips for helping? Currently, we have different sections per role, but I'm thinking that for some of the topics, maybe it makes sense to have one single article that addresses each role type, and AI can mention that. “If you are a [basic end user], then you will need to reach out to your [manager]. But if you are a [manager], then do this: ______” Other ideas?

It's kinda absurd that in 2025, Zendesk doesn't have an easier way to automate CC-ing external parties without the need for webhooks.   Add a trigger that has conditions that target an End User or Organization and let me cc someone on the ticket.

Zendesk's AI Agent Tickets allow agents to view tickets before the Bot hands off to an Agent, but sometimes it's clear that an agent should jump in and offer guided assistance to the customer.  This is a planned feature for a future release, but it can be implemented now with the App Builder. Problem StatementAgents, Team Leads, and Admins need to be able to take control of AI Agent tickets and turn them into agent tickets so that they can reply to the end user.  This can be helpful if the agents have more flexibility with policy than the AI is allowed.   As long as the Bot has control of the ticket, the agent can't assign or comment on the ticket.  When an AI Agent is transferring to an agent, the bot will prompt the end user to confirm that they want to speak with an agent.   The app we're building will take the AI Agent ticket and reassign the conversation to the default group in Agent Workspace.  The ticket can then  be reassigned t

Hi there, I hope you can help. I have set up an automation for requesters to get a follow up email. I've used tags in order to trigger the follow up email as well as another to stop it from firing more than once. My issue is that as the condition for the automation to work, the ticket needs to be marked as solved. So when requesters reply to the follow up email, the ticket status is not changing to open despite a reply. The ticket status is staying as solved. I need to get the ticket to reopen if requesters reply to the follow up email, If there any action I can add to ensure this happens? Looking forward to hearing from you. Best,  

Hi Everyone, Im helping our CRM configuring there Conversation bot/Chat bot on Checking for agent availability. He provided a link to me as her guide on setting up. https://internalnote.com/a-full-guide-to-handling-agent-availability-in-zendesk-messaging/#checking-for-agent-availability  On article it said this the endpoint.GET 'https://domain.zendesk.com/api/v2/agent_availabilities?filter[channel_status]=messaging%3Aonline'  But when i try it. Im receiving an Error 401 which is related to authentication.  

I'd like to scan our end users for identities that are undeliverable so that we can automate rudimentary bounce error reporting. The only method I can think of is to loop through each end user, then loop through their undeliverable count / deliverable value. Is there a way to scan/filter for identities that have been marked undeliverable or that have an undeliverable_count > 0? 

Currently, all tickets created through the chat/contact form come in with the same default subject:"Troubleshoot a technical issue (error messages or crashes)" This makes it difficult to quickly identify and differentiate tickets in the queue. Introduce automatic subject generation for tickets, ideally by: Allowing AI or rule-based logic to suggest a ticket subject from the initial issue description. Including subtype names and key error terms in the subject line. Keeping the option for customers to manually edit/override the subject if needed.

We're currently implementing AI Agents Advanced and I am curious if anybody has any instructions they use that have been useful in tailoring the agents to their specific needs? I'd be interested in instructions related to general customer support best practices or those that address a specific need. 

Dear Zendesk and Community,  I searched the whole Admin Help Center several times by now, and shockingly have not found anything on this topic?  We really need a way to export conversation transcripts from the bot insights. Or at the very least a report of the initial visitor requests, as in the first text entries, in bot conversations for the past 90 days. It is highly inconvenient that we can only see the conversation insights for the past thirty days manually in amdin center, and there is no way to export that in a report? Did I miss an important update and overseeing something? If anyone knows how to do this, I would greatly appreciate any advice here. IF this is indeed still not possible, dear Zendesk Team, please let us know when we can expect a feature like this to be released. I am pretty sure everyone that is using the native Zendesk Messaging Bot would need this.  Kind regards,Yvonne 

Tried creating a trigger to autoclose solved tickets if a user responds to say thanks, would not work with “Misc::Thanks::Thank you” intent as it is only available for AnswerBot and Messaging channels. Zendesk support advised they are developing an Intent dynamic prediction, which allows update the intent based on the most recent messages, so this would allow us to perform the close solved tickets if user replies with thanks use case, however this is not available yet and no ETA. Any further information on when this will be available on email channel would be appreciated.

Hello, There is documentation on how access to knowledge base articles can be restricted per user, but how do these user permissions interact with the different AI features in Zendesk? For each AI feature (e.g.  chatbot answers/article suggestion, procedures for auto assist):Are articles which the user has no access to downright ignored during processing?Or is there be instead room for information leaks (note that we count stating that an article exists as a potential information leak as well)? Moreover, what should we do about cases where AI actions should be informed/conditioned by knowledge base articles to which the requesting user has no access? Any recommendations?

I understand that the Zendesk bot generates answers based on the Zendesk Guide knowledge base. Specifically, I am interested in knowing whether the Zendesk Advanced Bot can search for content from specific pages outside the Zendesk Guide, such as our main website. Furthermore, I am curious to know if the bot can recognize content within images on item pages. 

Wondering if anyone using the Generative AI bot in their help centre has experienced their bot providing incorrect article suggestions and if so, what is the customer experience?  We are working on implementing this in a couple of our help centres and have to ask other users what their experience is for our approval process.  Any feedback based on your expereince would be appreciated.  Thanks

Hi Krzysztof Dobrzyński  In this article you mention roll-out ends on August 7, 2025, is that correct ? Because my Zendesk subdomain doesn't have the feature activated yet. On other hand, I work with a client who has it activated, but doesn't have Chat nor Messaging (Legacy version Support Professional + Explore and Guide Professional plan). Can we setup an AI Agent for Email / Web Form without having Messaging activated in the account ?Thank you

We integrated the AI bot with our telephony service. Our aim is to automatically retrieve any entity saved on the bot's session (e.g. a customer name, call reason, etc.). We are of course able to send any kind of entity or result via a pre-defined dialogue or process as an API call. But this is complicated, customers need to know which variables are being gathered in the flow and it takes time to set up. Therefore, this isn't what we are looking for. We would like to automate this process. Ideally, we receive the bot's answer together with the available sessions via the webhook. Or we have an entpoint that allows us to retrieve all available sessions. This way we can further work with this data in the next part of the call interaction. We were made aware that the more or less retired Action feature could be a solution here but that it isn't being developed further. Looking forward to your answer.

We utilize a messaging AI Agent with business hours as 7AM-4PM. Within the AI's configured “Answers” we have a condition at the top to flow to a set of answers depending on what time the chat was initiated. During open, most chats end up being routed to a ticket. I'm running into an issue where chats are being initiated at 3:59PM for example but the user doesn't make it to the step in the flow to create a ticket until 4:02PM at which point our support staff is gone for the day. Support has confirmed that the AI Agent doesn't have a way to natively be aware of the business hours status changing mid-chat. If the Answers flow builder allowed me to do something like “route to step *" or “Restart Answers”, I'd be able to insert a business hours check before our ticket creation step and route these back to the beginning to be handled by our main business hours condition.  Does anyone have a suggestion for working around this issue? 

Good day! Has anyone with a non-zendesk knowlegde base portal implemented the AI Agent for Messaging bot on their product / platform? I am curious to find out:CONSIDERATIONS: Your organization has a web based product + a 3rd party Help / Knowledge Base platform + the bot being installed on the product. Your observations leading to the Go-Live Has volume increased after implementing the bot on your product / platform? Are users bypassing the knowledge base to create tickets with your support team? Have you seen this as a temporary phase and does the volume continue to rise (OR) has this normalised to pre-bot period?

I need help setting this up for my site. I have already created a comprehensive help centre with FAQs and answers for the most commonly asked questions. Is there a way to automatically scan the FAQ pages and make the questions, prompts and answers for the AI Z bot? Or does all of this have to be done manually? If there's no automatic way, is there a third-party bot that can scan a FAQ page and create the bot questions, prompts and answers. Thank you.

I am writing an automation to accommodate for rate limit errors - specifically 429s. In order to do so, my error handling needs to pull the [ratelimit-reset] property from the error message, then add a delay. However, without hammering the end point and causing whatever errors across the production API, I'm not able to test for real world results. Is there a way to test where you can manually trigger a 429 message without actually causing one?

Hi all! I know Zendesk AI can suggest many object creations like macros and triggers. Can Zendesk AI also suggest views or make views on the fly? I know this is where we're going with Explore reports & dashboards. I am wondering the same if AI has this capability for views too.  I hope one day you can interact with every Zendesk object this way and prompt your way to glory; or the model knows what to build next for you with 1-button build actions. If anyone can shed some light on the AI View Builder topic, please let me know. Thanks in advanced :)

Our software utilizes maps and we often receive requests to update the GIS data (i.e. this street name changed. this steer is bi-directional on the map, but it is a one-way street). We'd like to create automation around these requests, so I created a custom intent. However, we also receive many other requests related to things that involve plotting addresses, routes, stops, etc. The custom intent is understandably being applied to both types of requests despite my attempts to include exclusionary statements in the intent description.  Has anybody found an approach to making custom intents distinguish between requests of different types that may use similar keywords or phrases? 

Hello,I reviewed my billing and noticed that I was charged for something I do not use nor enable: Automated Resolutions Usage.I would like to know how to disable it. Regards,John