
- Post History
- Subscribe to RSS Feed
- Mark as New
- Mark as Read
- Bookmark
- Subscribe
- Printer Friendly Page
- Report Inappropriate Content
on 03-29-2021 06:13 PM
The 2019 Attivio acquisition has been re-platformed into the Quebec code base and it brings a new “consumer like search” experience to our customers. Attivio was the market leader in cognitive search using AI capabilities such as Natural Language Processing and Machine Learning to enhance search. We’ve taken the best of Attivio and the best of our platform capabilities and created the AI search product. Customers who upgrade to Quebec will have a choice to stay on Zing or move to AI Search.
In this article I’ll speak to three key reasons why you should consider switching to AI Search.
Reason 1 – A Consumer Like Search Experience
As consumers we expect a Google or Amazon search experience where we are presented with personalized relevant search results. AI Search brings that consumer like search experience using Natural Language Processing to predict the users search intent and takes into account how users searched in the past to return the most relevant search results for the user.
So when a user enters “iPhone” (figure 1) we return the top result as a Genius Result (1). The Genius Result returns the search object with the highest relevancy score for our search. The relevancy algorithm prioritizes catalog items over knowledge articles as the intent is to order an iPhone. AI search uses a search result improvement rule to promote the latest iPhones in our catalog (in this case it’s an iPhone6 - trust me my catalog has iPhone4 & 5's in it). In addition, the Genius result includes an ability to take action on the search; in this case the action is a request button (2) which can use any API on the ServiceNow platform to kick off a workflow. Finally, the user can click on the filter facets to include or exclude results from different search sources (3).
(figure 1) Genius Card and Top results
Let’s break down the iPhone search from the developers Search Preview (figure 2). For starters our relevancy engine is predicting the relevancy scores (1) to determine what order to return the search results. Relevancy scoring also takes into account the end user’s past interactions with AI Search. In (2) we executed the “Deprecated iPhones” improvement rule (2) to only show the latest iPhones in our catalog and our NLU intent matching predicted the search intent to be “catalog find” (3).
(figure 2) Genius Results relevancy ranking
Reason 2 – Search any Source
I’ve been in many meetings where my customer will say “we have our knowledge articles not just in ServiceNow but in SharePoint and other sources, can you index and search those as well?”
Absolutely, AI Search enables the platform to index and search external sources. In (figure 3), we can index and search any table within ServiceNow. AI Search will leverage integration hub to natively deliver integration to SharePoint Online in Quebec with more integrations to follow. If you need to integrate to additional external sources such as Adobe, Atlassian, IBM, Oracle, SalesForce, MSFT Teams we have a partner that has built 45+ AI search connectors that can be used in a paid engagement. Notice AI search can be used in mobile, portal, and Virtual Agent.
(figure 3) AI Search architecture
Once you’ve integrated to your search sources AI Search leverages the Entity View Action Mapper (EVAM) to control how to display the search results from multiple data sources back to the user, see (figure 4). EVAM helps us grab snippets of knowledge articles that are the most helpful and displays them back to the end user, so they don’t have to read the entire knowledge article.
(figure4) – Entity View Action Mapper
Reason 3 – Actionable Search
Because AI Search is part of the platform, we can use the search results to deflect incidents before end users create a ticket. AI Search Assist uses AI search on the Service Portal and ServiceNow Mobile to help with incident deflection. (Figure 5) shows a rendering of what this will look like on mobile. AI Search Assist is part of the incident record producer and when configured provides search results with a button to take you to a catalog item to resolve your issue without raising a ticket. If you’d like to learn more about AI Search please reference our Quebec documentation.
(figure5) – AI Search Assist
- 5,772 Views

- Mark as Read
- Mark as New
- Bookmark
- Permalink
- Report Inappropriate Content
As I mentioned in the article, ServiceNow will initially release a MSFT Sharepoint connector. If you need to move to AI Search today and connect to other external sources please see the below for more details on the 45+ connectors to AI Search.
AI Search + Raytion enables our customers to integrate 45+ third-party information systems in ServiceNow applications and deliver a truly unified search experience.
https://www.raytion.com/availability-servicenow-ai-connector.html
- Mark as Read
- Mark as New
- Bookmark
- Permalink
- Report Inappropriate Content
This is great, Lener, thank you.
Another partner that has seen success in this area is Coveo Software Corp. Their Standalone Application on the store “AI-Powered Search & Knowledge Recommendations by Coveo” is ‘Built on Now’ meaning the entire end user experience lives on the ServiceNow Platform and customers can use the application without the need of leveraging IntegrationHUB licenses to perform off platform searches.
Coveo for ServiceNow - https://www.coveo.com/en/products/integrations/coveo-for-servicenow
Coveo Docs - https://docs.coveo.com/en/1479/coveo-for-servicenow/coveo-for-servicenow

- Mark as Read
- Mark as New
- Bookmark
- Permalink
- Report Inappropriate Content
Hi Lener,
I am try enable it on PDI but its not working it showing me like these please help me out

- Mark as Read
- Mark as New
- Bookmark
- Permalink
- Report Inappropriate Content
Hi Samyak,
Unfortunately at this time we don't support PDI's for AI Search.
AI Search can be enabled for sub-production environments.
Best Regards, Lener

- Mark as Read
- Mark as New
- Bookmark
- Permalink
- Report Inappropriate Content
Hi Lener,
In Dev instance also it showing same status.

- Mark as Read
- Mark as New
- Bookmark
- Permalink
- Report Inappropriate Content
Hi Samyak,
On dev please try out the troubleshooting guidance on KB0867751. I would recommend opening a Hi ticket as support will be able to configure your Dev instance to point to the predictive intelligence infrastructure used by AI Search. -Lener

- Mark as Read
- Mark as New
- Bookmark
- Permalink
- Report Inappropriate Content
Hi Samyak, unfortunately we do not support PDI's at this time as AI Search uses the Predictive Intelligence infrastructure. Only customer sub prods and prod instances are enabled with AI Search.

- Mark as Read
- Mark as New
- Bookmark
- Permalink
- Report Inappropriate Content
For customers using Virtual Agent on Quebec there are out of the box VA conversations that use AI Search.
Quick summary of AI Search on VA

- Mark as Read
- Mark as New
- Bookmark
- Permalink
- Report Inappropriate Content
My peer, Chris Steinke, compiled the below K21 AI Search enablement material that can help you get started on AI Search. -Lener
AI Search Enablement:
- AI Search Success Meetup session
- Empower users with consumer-grade, AI powered search experiences
- K21 Get Started with AI Search
- CCL1084-K21 – Customizing AI Search

- Mark as Read
- Mark as New
- Bookmark
- Permalink
- Report Inappropriate Content
Hi Lener,
Question for you - is there any functionality we lose, or downsides to moving from Zing to AI Search?
Currenly the search log table stores all seached, even when the property 'glide.service_portal.search_as_you_type_behavior' is set to 'typeahead' (which means when typing in the /sp portal the results are shown straight away in the dropdown, not needing to press enter).
Keen to understand the landscape before we recommend to our customers.
Thanks
Carl.
- Mark as Read
- Mark as New
- Bookmark
- Permalink
- Report Inappropriate Content
You might want to look at following documentation for your question:
https://docs.servicenow.com/bundle/rome-platform-administration/page/administer/search-administration/concept/search-suggestions.html
Mark this helpful/correct if it works for you
-Anu

- Mark as Read
- Mark as New
- Bookmark
- Permalink
- Report Inappropriate Content
I often get asked what does AI Search use for it's relevancy ranking and I wanted to point our community to Heather Phipps' excellent community article on Relevancy for AI Search.
Here is an excerpt from her article...In Quebec and Rome, AI Search relies on a variety of content features for relevancy ranking of search result documents. These include:
- Document freshness — how recently the document was created and/or updated
- Document popularity — how often the document has been viewed
- Title match — how well the query matches the document’s title
- Content match — how well the query matches the document’s body content
- Tags match — how well the query matches the document’s tags
- Tags popularity — the popularity of the document’s tags
- Mark as Read
- Mark as New
- Bookmark
- Permalink
- Report Inappropriate Content
@Lener Pacania1 we have below 2 filter options Last modified and View Count OOO in Zing search which is not present in AI search. How can we create similar filter in AI search for knowledge Bases.
- Mark as Read
- Mark as New
- Bookmark
- Permalink
- Report Inappropriate Content
Hi @Lener Pacania1, I have the same question as @Gunashekar. Is there a way to add the last modified and view count?
Also, is there a way to apply multiple filters for example articles in Knowledge base 1 OR Knowledge base 2?
- Mark as Read
- Mark as New
- Bookmark
- Permalink
- Report Inappropriate Content
Within the search application configuration you can add facet AI Search > Search Experience > Search Application. Once in the appropriate application select the 'Facet' tab and you can add a new 'Last Modified' facet with the kb_knowledge.sys_updated_on field (in Utah you could add multiple table into the same facet as well shown below). You can do the same for view count however to group results into ranges it would require some additional customization.
- Mark as Read
- Mark as New
- Bookmark
- Permalink
- Report Inappropriate Content
WOW... There are some seriously smart people working at ServiceNow.