Recent searches
No recent searches

Stuart French
Joined Aug 23, 2024
·
Last activity Feb 27, 2025
Following
0
Followers
0
Total activity
30
Votes
14
Subscriptions
6
ACTIVITY OVERVIEW
BADGES
ARTICLES
POSTS
COMMUNITY COMMENTS
ARTICLE COMMENTS
ACTIVITY OVERVIEW
Latest activity by Stuart French
Stuart French created a post,
Background info:
Ultimate bots (recently renamed to Zendesk ‘AI Agents Advanced’) are unable to handle relative date questions without being prompted first on what the current date is.
For example, if a Zendesk knowledge-base article lists the release date of various products, and a customer asks "is Product A released yet?" or “Will Product A be released this month?” , Ultimate cannot answer the question. Likewise, if an internal article lists a series of dates that agents are ‘On Call’, & a user asks "who is on call today?" or “Who is on call this coming weekend?”, then the Bot is unable to answer.
If the user first states the current date at the start of their question, then the Ultimate bot is able to answer relative date questions. Of course, the user may not know that they need to state the current date before asking their question.
Ultimate bots did used to have this option:
‘UltimateGPT> Bot Persona> Allow answering outside knowledge base'
However, using that option made no difference. The feature ‘Allow answering outside knowledge base' seems to have disappeared now.
We are told (in ticket 13337791) that currently there is no in-built feature for Ultimate to be aware of the current date, and that we must integrate a Date API or develop a custom script that fetches the current date. However, Zendesk Support cannot support or help with any custom scripts or APIs.
The feature request:
This feature request is to ask for an in-built feature within Ultimate (Zendesk ‘AI Agents Advanced’), so that it is automatically aware of the current date ‘out of the box’.
Posted Feb 24, 2025 · Stuart French
0
Followers
1
Vote
1
Comment
Stuart French created a post,
The feature being requested:
We are testing Ultimate bots (recently renamed to Zendesk ‘AI Agents Advanced’). We are using the option Generative AI with all options selected (including ‘Activate zero-shot interface’ and ‘User zero-shot AI model’).
We find that sometimes it gives inaccurate or misleading answers. We would like a way to train the Model to improve specific answers.
Background information and use case:
I realise that we can create ‘Use cases’ to push certain questions down a specific path towards documentation that give the right answer, but it would be good to be able to train the bot in circumstances where it has given a specific bad answer.
We can access ‘Conversation logs’, and in there we can 'Provide feedback'. When we do that we can mark the answer as one of the following:
- Correct/well answered
- Incomplete answer
- Incorrect article used
- Made-up facts
- Incorrect language or vocabulary
- Other
It would be fantastic if the above feature actually trained the model to avoid the answer that was given. However, that is not ‘training the model’. Instead it is described as “This feedback is analyzed internally to improve UltimateGPT’s performance”. Zendesk Support (in ticket 13343928) confirmed that this option is not used for training of the bot specifically, but more to 'help Zendesk's backend developers train the overall models'.
Currently there is no feature available to tell the bot that a specific answer is bad (ie. train the model). Using ‘Use Cases’ to guide the bot for topics that lead to frequent bad answers is not really feasible, as there are many different scenarios where the answer is wrong, or slightly wrong.
Posted Feb 20, 2025 · Stuart French
3
Followers
1
Vote
1
Comment
Stuart French created a post,
I have set up various UltimateGPT Web Crawler Knowledge Sources. Most work OK. However, I have set up one that points to a URL that presentsa PDF file (the URL ends with the name of a PDF file)
It builds successfully, however in the Import Summary we see that Results was 0 and it extracted no text.
Testing the bot, asking it various questions about things that are mentioned at that URL always results in the message 'Sorry I didn't quite understand'.
Following discussions with Zendesk support in ticket 13337512, it has been concluded that the UltimateGPT ‘Web Crawler’ Knowledge Source is only supported with HTML pages.
At https://support.zendesk.com/hc/en-us/articles/8842717836442-Unlocking-AI-Success-with-Zendesk-Knowledge-Base-Event-Summary in January 2025 we were told:
"Currently, Zendesk's generative AI responses utilize only text-based content. This includes information from articles, community posts, and federated sources when using our federated search. We’re excited to share that support for embedded graphics, images, and text files (including PDFs) is on our roadmap and will be available later this year!"
So, this feature request is to ask for UltimateGPT ‘Web Crawler’ Knowledge Sources to be able to crawl and use URLs that point to PDF files.
Posted Feb 19, 2025 · Stuart French
3
Followers
1
Vote
1
Comment
Stuart French created a post,
We would like to request a supported method to have an ‘Internal’ section to the bottom of articles that are visible to customers. So, for articles that are visible to ‘Signed-in Users’, we would like to be able to have a section at the bottom of the article which is only visible to ‘Agents and admins’.
The use case is that it is convenient to be able to add notes for internal viewers that provides further background on the subject matter, like the ticket that an article was spawned from, or the Jira that it is linked to, or any further internal information that is relevant. Internal Notes would be a reference to help internal agile communication.
Presently, we have a customized method that we use to do this. Internal Comments can be added to our articles using a simple Custom CSS and JavaScript. The Internal section is seen when viewed by an Admin/Agent, but is not seen when viewed by a Signed-in User (a customer).
The trouble is that when we implement Ultimate Bots (recently renamed to Zendesk ‘AI Agents Advanced’), then although the internal text is not seen by the Signed-in User, the Ultimate bot answer does provide information from that internal section to the user. We don't want this to happen.
After discussing this with Zendesk support in ticket 13337690, we have been told that our ‘Internal section method’ is out of scope and an Ultimate Bot limitation, as the bot does not have a way of knowing what it should/should not tell the enduser(requester).
So our Feature Request is two fold:
1) Provide the ability to insert an ‘internal section’ to the bottom of Signed-in Users articles, that is viewable to Agents/Admins
2) Ensure that Ultimate Bots ( Zendesk ‘AI Agents Advanced’) do not make suggestions to Signed-in users, based on the information in the internal section
Workaround until the feature is implemented:
- Have two articles. One containing Signed-in user visible info, and a second article that holds the internal information and references the public article.
Posted Feb 19, 2025 · Stuart French
3
Followers
2
Votes
2
Comments
Stuart French created a post,
We use Omni Channel Routing to automatically route Email and Web Form tickets. It works well.
However, Omni Channel Routing currently does not handle tickets that are created within the 'Zendesk Support' app (which are created as Channel 'Mobile', and end up displaying ‘via mobile’, instead of ‘via email’ or ‘via web form’). These tickets do not get assigned to an Omni ‘Queue’ and therefore do not get auto assigned to an Agent. We have to manually assign these tickets to our Agents.
It has been confirmed to me in Zendesk ticket 13311250 that the channel of Mobile
is not listed in the Omni documentation (at https://support.zendesk.com/hc/en-us/articles/4409149119514-About-omnichannel-routing) as a channel that would work. Therefore, even if ‘Zendesk Support’ is an app produced by Zendesk, Omni Channel Routing is not currently able to handle these tickets.
The use case is that our Light Agents are sometimes ‘on the move’ and it is much easier for them to use the ‘Zendesk Support’ app to create an urgent ticket (& then update/view that ticket), than it is to use the alternative of ‘email’.
Therefore, we would like to request that Omni Channel Routing be enhanced to also be able to route ‘Mobile’ tickets that are created via the 'Zendesk Support' app.
Thanks.
Stuart
Posted Feb 14, 2025 · Stuart French
2
Followers
1
Vote
1
Comment
Stuart French created a post,
Our Admin users are using an Explore Dashboard to view and change the Availablility status of our Agents. Using this method we can Bulk change some of our Agents at the same time. Or they can change the status of other individual agents. Non Admins can use this to view the Availability of other agents, but they cannot change their Availability.
I also see that post https://support.zendesk.com/hc/en-us/articles/8513847999514-Announcing-new-bulk-APIs-for-agent-availability-with-omnichannel-routing has recently been created advising how bulk agent availability changes can be done by Admins via API.
We have some Agents that are ‘Team Leads’ who have responsibility for a monitoring and changing other Agent availability. However, they are not Zendesk Admins.
We would like to request the ability to specify a group of specific non Admins to have to ability to do this please. We don't want to give Admin access to these ‘Team Lead’ agents, we just want to give the ability of changing other Agents Availability status's. This would allow our Team Leads to be able to make these changes, instead of having the chase up the Agent manually, or ask an Admin to do it for the Lead.
Thanks
Posted Feb 05, 2025 · Stuart French
2
Followers
1
Vote
1
Comment
Stuart French commented,
Adding a comment because I see this ticket as ‘Awaiting your reply’, but I have no further updates for this. Can you please set this to an alternate status?
View comment · Posted Jan 21, 2025 · Stuart French
0
Followers
0
Votes
0
Comments
Stuart French commented,
Thank you. Do I need to do anything else in this ticket?
View comment · Posted Dec 19, 2024 · Stuart French
0
Followers
0
Votes
0
Comments
Stuart French created a post,
I have created a View that lists certain Support tickets. I want to be able to create a report which shows the 'Intelligence> Summarize Conversation' for each of the tickets listed in the View. We can do this manually by copying/pasting the 'Summary' from the Intelligence pane of each support ticket. But we would like to be able to do this automatically (in one go) for a list of tickets in a 'View'.
We are asking for this functionality because we love the generative AI ‘Summary’ (Summarize Conversation) feature. An example usage is for an accout manager who wants to display currrently Summaries for all open tickets for a specific customer organization/account. It would be fantastic if it was possible to list within Zendesk (or export) the tickets for that organization together with the ‘Summary’ info for each ticket.
I raised this in a ticket with Zendesk (13104062) and it was confirmed that this is not currently possible to accomplish through Explore. There is also no current way to extract this information via the API as a workaround. The Zendesk Engineer said that ‘this would make a great product feedback request in our community forums’. Hence I have raised this here.
Posted Nov 21, 2024 · Stuart French
3
Followers
5
Votes
5
Comments
Stuart French created a post,
We would like to verify some articles every two years and others every three years. At present, when setting up an article Verification Rule, the maximum frequency that can be set is ‘1 Year Interval’. We would like to be able to set a longer period of time, 2 years or 3 years.
Additionally, during the article review process there is no way to automatically send a reminder for articles that are ‘Awaiting Review’ (to the assigned reviewer), or for articles that are ‘In Progress’ (to the author, as a reminder to finish creating their article). We thought of setting up a Verification rule for the folder that holds our draft articles. However the minimum Frequency for Verification Rules is ‘2 week interval’. We would like to be able to set that to a lower period of time, like every 3 days or every week. But that is not possible.
Ideally, it would be good to be able too set the frequency independently (so we can set the period of time ourselves). But if that is not possible, then please add further options of 3 Days, 1 Week, 2 years and 3 years.
Thankyou.
Edited Aug 23, 2024 · Stuart French
2
Followers
1
Vote
0
Comments