Employee

Ask Us Anything: about TOPdesk's AI features

4 replies
61 views
Translate All

Ask us anything about AI in TOPdesk. Until 6 November, our product managers, customer success managers and other colleagues are here to answer your questions.

Wondering how a feature works, how it fits your setup, or what to think about before you switch it on? Ask away. Some features to get you started: TOPdesk Robin, AI Triage, MCP and the AI Readiness Dashboard. You’ll find the full overview here.

How to ask

Post your question in the comments below, one question per comment. Read along too, because someone may have asked what you were about to.

We’ll answer throughout the week and tell you what we know today. If we don’t have an answer yet, we’ll say so.

So, what would you like to know?

AI

NickB

Contributor

Within Functional Settings > General > Search terms used in Self-Service Portal

i can download all the search terms used by customers using the SSP.

Would it not be beneficial to have a column within the results to determine Ai searches vs normal to ascertain adoption or potential issues customer feel the ‘classic’ search might use better (or just avoidance of AI) and i hope to see a visible different of keyword searches (classic) vs question / conversation (ai search)

Probably one for TIP i'm guessing; but useful insight none the less and will probably turn into more of a dashboard request (see below)

I’ve recently exported these search terms to aid insight and a round of keyword meta data fixing before going all in on the AI-SSP to give it its best chance.

its found 65,518 searches with 14,594 unique terms with 17,735 rapid exact repeat search terms.
36,273 are one word search queries.

i’ve created a little graph showing use of search terms per hour over the time span and graphed out the top search terms and categorised/grouped them by theme. and it matches our service catalogue structure pretty well which reinforces the design choices we made (or did we drive the search to fit it?)

I’ve spat out another view for acronyms and synonym opportunities looking at spelling / typo’s / preferred or higher matching terminology vs lower use / variants of the same thing and count of them and any recommended actions (this was all done with a bit of Ai trickery too)

i’ve then asked Ai to prioritise the opportunities (28 of them and 10 high priority) and its certainly going to be useful.

This feels like a potential AI request in itself to create said dashboard or equiv.
What more value do you (and others reading this) feel this would add to the product?

Translate

All Round IT Nice Guy + "Guardian of Good Enough"

Simon Rupp

Employee

Hi Nick,

Thanks for sharing this.

One thing to clarify on the AI vs. classic column: when AI search is enabled in the new SSP, there is no real separate "classic" search anymore. So a flag to tell them apart wouldn't be possible after the switch.

The need behind your idea is still a valid one: understanding whether search works for your users and where it doesn't. Beyond the search terms themselves, I'd expect the most valuable signals to be:

- searches that return no result and what were the keywords (aggregated) that didn't return any search results?
- whether users click a result after searching (if they don't, that might be a proxy for the search having returned only poor results - However, with our AI summary of the knowledge items found, this may also be a misleading proxy when it comes to FAQs)

Your synonym and acronym analysis points in the same direction: not just seeing the data, but being able to act on it.

I'd really encourage you to submit this as an idea on TIP, ideally describing also a dashboard you have in mind and the decisions it would help you make. The more concrete the use case, the better we can assess it.

I'd also like to hear from you, Nick, and from other Community members: which of the signals above would matter most to you? And what outcome would you want from them? For example, recognizing that you need more service offerings, better service descriptions, or more knowledge items, so that there is always a good answer to every question.

Thanks again for sharing your findings.

Simon

Translate

NickB

Contributor

I can see on TIP things like AI clustering of Issues that are becoming items to exist in a problem management module space - should something exist for major incidents?

Detect a developing major incident, group related calls, identify affected services, draft comms and maintain a timeline. extending to strong suggestions that "There are already 37 calls that look like this. Link this to Major Incident MI-123." type of operator aid too.

What about a CSI-Ai aid
"Password-reset demand rose 31%. Here are the likely causes and three ways to reduce it."
Something that exists on the service creation pages to guide the design of the service - this could even point at making knowledge more visible by suggesting items to be linked to a flow.

Do you think we’ll ever get a tool to Measure Ai’s deflection, minutes saved, resolution improvement and avoided ticket potential?

What about an Ai service manager for a human to interact with and bounce idea’s off of…
in my mind its something like

AI Service Manager │ ┌─────────────────┼─────────────────┐ │ │ │ Understand Decide Act │ │ │ Incidents Risk/SLA Workflows Knowledge Priority Integrations Assets Root cause Communications Changes Next action Ticket updates │ │ │ └─────────────────┼─────────────────┘ │ Learn & improve │ Problems / trends / knowledge

Translate

All Round IT Nice Guy + "Guardian of Good Enough"

MarloesM

Employee

Hey Nick,

I’ll respond to your interesting message one topic at a time,

On case clustering: We will start using case clustering in the existing answer assistant functionality. Currently answer suggestions are based on knowledge only, where often knowledge is still only captured in historic incidents. Using those when knowledge is not available, will make sure knowledge from your TOPdesk environment is utilized more efficiently.

Moving forward, case clustering indeed has the goal of making sure bigger issues like majors or problems can be spotted proactively, this is the direction we want AI in TOPdesk to move into.

For incident deflection, like the password example, TOPdesk indeed needs to know what processes are in place at an organisation which required guidance by for example the service manager. This is something we are actively exploring and I will tag my colleague Marvin Pieper to contact you about that!

Translate

Marloes Meijer, Product Manager TOPdesk

How to ask

Sign up

Login with your MyTOPdesk account

Login to the community

Login with your MyTOPdesk account