we are will be on release the features um so we'll need to share the Safe Harbor load notice first all right so let's jump right into our overview so today our topic is AI search for next experience but before we jump into the content there I do want to start our presentation with a little bit of the history and context of AI search so I believe some of the folks on the call today are already familiar with AI search or study in Quebec AI search has been embedded into the servicenow platform to support what we call requester use cases or self-service use cases so some of you may have already used AI search in Virtual agent portal and mobile and then starting in Tokyo time frame uh we are extending AI search support into the fulfiller and agent or admin type of personas so today's topic we'll be focusing on the global search and configurable workspaces as the search interfaces for our admin users fulfillers and agents so for folks who just started to learn AI search um because like we're extending up into new use cases don't worry uh in today's session we will be covering some of the fundamentals of AI search and then we will also be sharing resources all right so let's start with AI search overview so before we talk about AI search for next experience or as searching Global searchable workspaces I do want to emphasize that AI search is the platform product uh in servicenow which means that you know all the different search interfaces no matter service portal virtual agent or workspaces they will be able to take benefits of all the AI search K uh platform functionalities that we'll be covering right now so when looking at AI search or from product perspective we have three pillars for AI search one natural we want a research to be able to answer any questions in any different format across of course multiple languages secondly AI search should be smart so not only AI should provide a better relevancy out of the box we also want AI search to be even smarter so with that we have machine learning based relevancy embedded into our relevancy model to actually learn from our different interfaces users to understand how we can provide more relevant more personalized with results to our end users no matter your agent who are your AR request or employee lastly easy is definitely another pillar for AI search we're not only talking about you know having the search experience to be easy to help our search users to find the results it is also about how easy it is to set up AI search which I will definitely talk about in the later slides and with those three pillars we also have some of the um some of the goals that we have set for AI search for example first AI search should be self-learning we are using our AI Technologies again to automatically learn from our end users interactions with search and be even smarter setup shipping in minutes not month or we wanted to be closer and closer to this goal also with brick silos um not only breaking silos for searching service now contents that you can even bring in non-service now content into servicenows instance into AI searches or index and be able to service those uh like external content from AI search and lastly we also provide analytics we provide store apps that offers admin dashboard to show you how a asset performs we also provide a search analytics dashboard to show our business stakeholders on important Matrix about search such as average click position such as knowledge gaps so I'm going to quickly go over the uh The Mink key features or highlighted features I wanted to mention in each of the pillar uh so under natural we have a feature called autocomplete so for folks who are using Google every day you're probably already familiar with it which is when you are starting typing your queries from Google then you will be able to getting suggestions on queries or even results so to save your time on searching so AI search also offers the same capability when you start typing already we are already being able to find the Matched queries and results from your histories and even from the search results secondly we have the autocorrect spell checks which of course uh was not a feature available in Zing so our spell check is smart enough that we can understand the content in your index so even if there are some you know terminologies that you use inside your organization that can be detected as typos outside of your context AI search will be able to know that and we'll be able to know that you know even though that spelling is not common but it is the correct spelling in your organization thirdly we are working with a natural language understanding capabilities to help us understand the user intent and be able to provide answer and we use that feature in our genius results lastly we support multi-language so we support more than six languages for uh what we call uh Advanced Linguistics features uh such as spell checks synonyms so under smart we're going to be touched a lot on like the AI capabilities so the first thing machine learning based relevancy so as we know relevancy is the key uh measurement on determining whether a search engine is good or not good or Not So based on our actual customers usages uh we already know that AI search offers better out-of-the-box relevancy compared to a fine-tuned zing and then based on that I switch off also allow us to learn from our end users and then we have a machine learning relevancy model that learns behind the scenes and it automatically you know improve our relevancy to give even more relevant and uh personal uh personal personalized search results to our end user secondly we introduced the feature called genius results which is now available in uh portal and in Virtual agent or you will soon be available in uh in global search and workspace so GS results is that kind of like the answer box search experience that you know by detecting the intent being able to provide you the answer uh at the top suggested box so today we support three out of the box uh genius results the first one is question and answering which automatically extracts Snippets from your kdr KB articles if we find it a match in that uh snippet that we think can answer the search users questions we also offer the people genius results to help you find people and also a catalog item genius results and then lastly under smart we have a feature called result Improvement rules so we all know how you know AI could be you know smart and then you learn from our users however we also understand there are certain challenges on ai ai capabilities when we don't have enough data set between the uh between the to train the model so in that case some of the kind of human controlling becomes more important so with result Improvement rules as a admin user as the process owner you know your search use case you know your end user the best so you will be able to overwrite some of the relevancy rules or some of the results by using result Improvement rules to help boost promote or block certain documents based on the user context then based on the query lastly about search I'm sorry lastly about easy uh so for AI search we do offer a centralized AI search or admin experience so we will be touch touch base a little bit on that today as well in the demo so no matter you're you're trying to configure a asset for polo or what you're trying to configure AI search for workspace there is always the single place that you can go to and find all the AI such related configuration in there and also like we mentioned earlier it has such to provide search analytics capabilities which is available in one of our store apps that you can actually turn search usage insights into actionable Data can really help you improve your search uh search experience or even help you improve your knowledge base we also offer such quality evaluation tool which is a relatively new store app that we just released that can really help customers to understand you know how search how much better AI search is compared to Zing well compared to a different uh AI search relevancy model to really help you quantify that kind of like the value that AI search provides and lastly of course uh we will have we were providing a modern search ux and the intuitive navigation all right so with that I'm going to quickly talk about the high level architecture of AI search and as you can see today our Focus will be on global search and workspace so for AI search or AI search engine um on the left hand side those are the data that we can make searchable so of course you would imagine we can support searching servicenow data like KB articles catalog items task tables like incidents uh CSM cases so on so forth and another thing I wanted to mention is that AI search we also have the apis and the spoke available in integral integration Hub to support indexing external content so for example if you have a bunch of you know documents that you think will be useful for searching in SharePoint you will be able to index that SharePoint content into AI such as unified index and be able to surface those search results in those different search interfaces is there a question no no this is um I'm just going to remind everyone if there's any question please feel free to post them in the Q a section meanwhile I just want to remind everyone so regarding the now mobile portal and virtual agent boxes that you see here we did Cover how do you implement virtual uh AI in those experiences in our previous recording session so if you uh still have that link in the first of the chat you can go back to that Also regarding the external content we uh had a session on the I search for external content how do you structure that how do you build that it was a great session with Gerard I would strongly recommend that you look into it today we're just going to focus on the global search and the workspace but if you still even have any questions not regarding that about the assert we can take a look at it um go ahead Memphis awesome thank you Nabil great reminder all right so with that let's jump into our topic today um so AI search for next experience today is a store app and it is now available for all servicenow customers in the store and then it best best of all it's free so from the servicenow store you will be able to find AI search for next uh Experience Store app which would provide support to allow your Global search as well as your configurable workspace to use AI search so one prerequisite I do what search and workspace would do require uh your instance to be already have next experience turned on all right so with that I wanted to introduce some of the benefits especially for our folks who are new to AI search on the benefits When You migrate from Zing to AI search so first of all we covered some of the features uh that AI search offers and as uh if you're familiar with Zing you already know like a lot of them were not available in zinc so some of the key product features including this easy setup including the kind of like the additional configuration like query rules synonyms spell checks dot words can really help drive your uh search search quality and of course the different language support and all of course the lastly like the machine learning what AI powered relevancy to continuously improve the relevancy and lastly our AI search we also offer like analytics and then really help your organization understand how people use search in different applications and then so give you insights on how to improve them so when it really maps to the business outcomes uh as I mentioned Quebec is our first release so we are already seeing customer gaining values and then to have ai such to help them Reach their business kpis uh so things like having the faster resolution times or for improved experience and also really improve the self-serve experience in your portals and then really help deflect the cases and costs and lastly like being able to increase the agent abilities and also reduce the tuning needs and cost so those are the key kind of like benefits when you upgrade from zinc to AI search all right so last two slides before we jump into the demo so I wanted to quickly highlight some of the features that we're introduced in AI search support in global search and in workspace search and you will be able to get that from the store app that we mentioned earlier so first of all we will be bringing a unified and machine learning based relevancy across all different search sources regardless of where the data is coming from so if you recall how Zing used to look like all the search results are grouped so with zing in global search you often have to you know scroll all the way down to find the result groups uh to be able to find what you're looking for but for AI search we Rank by relevancy so no matter which data source the results is found as long as it's the most relevant documents you will be seeing them at the top also as you can see we are introducing a lot of new UI capabilities to really help agents and fulfillers to be able to quickly find the answers they're looking for which were not available in the old workspace word in the old Global search including the filters not only you can filter by the data sources you can also feel to buy additional metadata that's attached to your records things like you know States priorities categories and also the Sorting capabilities by default all the results will be sorted by relevancy but you can certainly configure other sorting options for example to sort results by popularity to sort results by updated dates and things like that and then best of all like all of those com uh all of those uh UI features on configurable so even if you have custom fields that you added to your tables that you are in your search to search on you can easily configure the fields in your customer table to be using that as facets or as sorting options and also another thing that I will be covering which is actually a newest features that we're introducing our version to release which is the release that we have released in November 3rd uh it's already available in the store is that in our version 2 of AI searching next experience we are introducing two new features uh and then we think it will definitely provide a lot of values to our customers or to use AI search index experience the first thing uh what we called is director list view so uh in search results we were actually introducing a button that allows you to direct your search queries from the global search results page into the list and form kind of like the into the list of this specific table so that's kind of like experience our customers are familiar with and of course like we know that in the list of view there's a lot of actions that you can do which you could not do in your in results page so the directors to be will really help you to quickly switch wheels and be able to take actions on those search results and then second major features that we introduced in version two is the migration pool so we have understanding that a lot of our customers you are already using zing in our workspace and Global search some of you may have already made configurations in your Zing experience so the migration tool that we introduced really helped providing the guided experience to help you guide I'm sorry to help you migrate the zinc configuration into AI search configuration and then have a easy flow to help you automating that you know that migration of the data sources from thing to AI search and then turn on AI search for your next experience that's great um if it's I think next is uh to go into the exercise and see this in action but uh well Memphis does that I would love to run uh Paul but you guys um I'm just gonna click launch let me know if you can see it I'm very interested in knowing that answer for the third question but I would love also to know for the first and the second so the first thing um what where do you describe yourself right now from implementation perspective when it comes to AI search um have you implemented it only for portal for workspaces or it's not applicable yet um and the second one uh if you didn't Implement AI search yet um what is the blocker are you looking to learn more is it a level of implementation effort the gaps and if there are gaps please feel free to write those answers in the chat would love to hear about it and yeah the last one what's your uh what do you expect for Thanksgiving dinner um here we're not in the US so where everyone is excited about the World Cup way more so our colleagues in back in the US and yeah the results reflect that so just give it one minute to let more people and be able to answer and again if you have more comments on the questions feel free to drop them in the chat foreign uh great results so as you can see people still planning on implementing and the the majority also people still need to learn more about it so for for that you we still have uh more sessions to come when we have search but we already released three previous uh on how to get started AI search on external sources and how AI search can help your uh and current implementation I highly recommend it to take a look at them and yeah the World Cup wins looking forward for the first game on Sunday uh Mavis back to you to show us how it works sorry I was on me thank you Nabil and then thank you guys for participating in the poll all right so uh let's look at the demo here so what you're seeing here is our brand new homepage of next experience and then in this instance AI search is already enabled so when you click into the AI search search box you will be already seeing some of the suggestions based on your previous searches so imagine a case that you wanted to just simply go back to what you have searched you can easily access it from here and then as you can see when I'm start typing uh we are already also suggesting based on your recent searches and based on your recent views uh to really save your time for typing one of the new features that we introduced in AI search is that we understand oftentimes our customers instance could have multiple workspaces associated with that so so does AI search we can also handle that so in the drop down menu you can also see AI search can support searching in different we call it like scope or different workspaces so let's do a quick search to see what the what the differences is oh I can type all right so when you do the search you will be getting directed to the search results page and then one thing you noticed is the uh we call it navigation tabs this navigation tab is actually automatically derived based on the search sources that you configured for the specific Search application so when you're doing the switch on switching from different Search scope or switching from different uh kind of like workspaces you can actually get a sense of what data sources they are actually searching on uh in the specific scope and really helps you make decisions on whether you are searching in the right place we need to change to somewhere else all right so let's jump into the real demo here um so here I'm going to start my search journey in uh the sou workspace I'm sorry the service operation workspace so imagine I'm a new hire in uh for uh for my organization and then somebody is asking me I to look into a specific instance and then in this case I get the incident number from my co-worker and then I think from our user research actually um we found this guy copy paste uh record number is actually the top usages of the global search today so as such certainly support that as well so when I copy and paste the incident number into the search box AI search not only can do exact match we also have the preview to help you understand oh is that something I'm really looking for and then no matter you can uh you can click on this record when you hit enter you will be directed to the record itself oh the geez I think they are this instance has been giving me issue this morning so this is all this is I this is the in real world this is why you will see that you know when you are doing kind of exact match when we are in the you know specific workspace that AI such of course will be directing you to that record in your existing workspace so in this case as me as a new hire in the I.T Department I'm seeing this person has a well-less access issue so I look into the incidents I couldn't find too much information for me to start working on however I do wonder you know if this uh issue happened before so in my search I started doing sound search Wi-Fi connection sorry about the performance seems like my might be my internet is giving me problem right now all right there we go okay so now I'm searching as you can see uh we are directly already default to search the uh service operation workspace because that's the workspace we're in currently and then when I do the Wi-Fi connection I actually fat fingered um so but it triggered AI searches spell chat functionality so AI search is actually able to direct uh able to detect that oh I have a spell spell uh typo in my query and they automatically correct my spelling now I'm searching the right spelling of Wi-Fi connection and also I can see the first thing I got directed to is the all tab so the all tab actually returns all the search results rendered by relevancy which means that the most relevant document will be ranked at the top no matter which data source it comes from and one thing that you will notice here is that you know AI such also uh is able to understand the synonyms we have a configuration called synonym dictionary that allow you to configure synonyms that's all you're using in your organization so in this case they understand okay Wi-Fi could be also uh referred as well list Network and we're also able to find you know incidents or documents uh which is using well it's all Wellness Network all right in my case um I'm actually looking for I know incidents I wanted to see how people solve the same problem before so I can then use my navigation Tab and then go into the incident tab so in this case all the search results I'm looking at are incidents and at the same time I noticed that I have additional filters that I can interact with oh so those are the metadatas from the incidence tables and then I can use that to further narrow down my search results so I know I'm looking only looking for you know incidents have assigned to network uh group and I'm also only looking for the incidents that has been closed so I wanted to see how the resolution was so in this case I can apply the facets clear uh easily and then quickly narrow down my search results and of course you can easily clear all the facets if you want yeah but uh with the uh with the facets I look at it like the uh the instance I was looking for and then from now I can just click into the incidents and then see how the resolution was and really helped me to solve uh my uh current open case and another thing I wanted to Showcase here is that you notice we have a little button here uh so again this is the unreleased feature so it might look a little bit different uh in the actual release but I wanted to give you a preview of this feature uh so when you click this feature it actually is able to direct you from our service um sorry from our results page into the list view so that you can actually see the Matched results uh using the same query and then from here of course we understand a lot of actions that you can do from the list of video that you can easily switch views and then do it from search results page sorry to interrupt you just want to be clear about it this is a feature that exists in Zing um or this new to AI search so this feature existed in Zing's Global search as well and then we carried it over as well so one difference actually on the bill uh it's a good question one difference I wanted to mention uh with AI search is that uh obviously like the search results page is using AI search which offers a lot more capabilities than what zinc offers however unfortunately the list view the search matching like the filter condition matching feature it doesn't have a sh support yet so in reality and then that's a important caveat for this feature in reality when you do the direction uh things were only carrying over the queries or the search results from search result page to the list view page might be different and also the ranking will be different as well because in of course in the list view it's not able to work by AI such as relevancy so that's some uh caveat that you wanted to clarify makes sense thank you awesome all right so I think oh sorry one more thing I forgot to mention uh the Sorting options so like I mentioned earlier the sortings are configurable by default it's sorted by relevancy so hopefully like the default ranking is already giving you what you're looking for and again just to mention again uh so we have machine learning based relevancy which means that for example if you do a query and then you keep clicking like the fifth results of that query then we understand okay for some reason and our machine relevancy model is actually trying to understand why a certain document is more relevant to this user to the specific query and with machine relevancy over time what you'll be seeing is that this fifth results will be ranked higher and higher uh and even to the top of the results when you do the same query in the future so that's how you will see the Machining uh relevance is actually working and learn from the end users and also just another quick note as mentioned our facet capabilities and our sorting capabilities are also configurable in the search application so that that is easily configured can be configured and you can definitely uh tailor it to or fit better for your use case so all right may this work while you continue with that I'll launch the second poll that we really want to understand your current uh status with their search and regard regarding the workspaces so what we're seeing today is how AI search can improve the workspace experience for agents and you know some of you might have one workspace or some they don't have it yet and but there is a caveat if you have more than one do you give your agents access to one only or more because maybe this this matters right do they need to switch the scope when they do the search foreign yeah so we wanted to understand like you know in your scenarios based on how you configure your access to your users for your workspaces is it the often case that you will you have a user that have access to multiple workspaces like the demo that we set up here or in your case it's more like you know one-to-one mapping yeah so this is really important for us to understand um and the other part is if you have done any customization to Zing um would love to know that and what kind of customization chart it seems was disabled we opened it now so if you have any notes that you would love to share with us would love to see it okay and the results are out so most people have um more than one workspace and most likely to let agents have access to more than one and it seems the majority don't have any customizations losing so migration might be easy right 30 minutes yes that would be easy that's good to hear great so let me resume sharing here um so okay so next topic I really wanted to quickly touch Point using our last five minutes is we're seeing how beautiful the new UI the new search experience look like now the question is how do I get there uh so here I summarized the three options for our customers to Levy 11 elevate your all next experience to AI search so three options I would summarize this kind of in the order of you know how easy it is so first option is that we do offer out of the box aash configuration for Global search and for I would say most of the workspaces that we work with so we do work with pus to understand the different use cases of the workspaces and then bu can package uh their own out of the box search configuration for that use uh for that workspace based on kind of the most common usages of that workspaces so the first option is definitely when you get AI search uh that you can take a look on what we have shipped or for your workspace as out of the box search configuration uh start using that and test with it and then you can always go back to the AI search admin admin UI to kind of fine tune it so I would say that's probably the easiest way for you to start trying out AI search fine tune it and then better cater it for your use case second option is to use our migration tool which I will show a little bit in the in the real demo so the migration tool actually again like mentioned it gives first it gives a very guided experience step one two three four uh to really guide you to the process to set up AI search for next experience secondly it has the automation to automatically migrate your Zing search sources into AI search search sources including configuring like you know the index sources including the triggering the index indexing automatically so you can certainly use the migration tool to migrate your existing zinc configuration to AI search so obviously this option is better fitted for those customers who have done more configurations on Zing so it will definitely help simplify the journey and then with that you know once the migration is done again you can start using AI search then in uh in next experience and then you can always again go back to fine-tune the search sources or other other features and then later on you know test and fine-tune same similar or similar thing and then the last I would say it's definitely caused maybe a little bit more effort is to manually set up AI search so there are three layers of AI search configuration of manual setup the first one is to Define what are the data sources that you wanted to index and then second one is to Define search sources where where you can Define some of the search behaviors like what data sources to be searched on with which type of filters defining the result Improvement rules or synonym dictionary and so on so forth and lastly the Search application layer allows you to Define UI presentations of your search page such as the autocomplete facets sorting so on so forth and then of course later on you can test and fun tune um so from here um I'm going to show you how the migration looks like of course our um demo instance has already been uh migrated to AI search so you are not uh you will not be seeing the um the process the actual process thing thank you but oh sorry this one jumped on me again can you see my uh migration tool not yet all right let me see how about now we can see it perfect all right so again we're now going through the process today but as I mentioned we introduced this migration to today's thing in the form uh but it does list out like step-by-step guidance and then he has the automation to help you um that we will actually automatically detect like which migration changes needed allow you to take a look review and accept and then once everything is accepted we will trigger the uh required searchable content and it will trigger indexing automatically and then as we know like indexing sometimes can take a while based on how much data you're indexing so we also have the notification system set up to actually being able to send you a notification once the indexing is done so once everything is completed then you can actually turn on AI search in your next experience directly in this form which was not available in the version one of the store app another useful thing that you can get in the migration tool is that we can actually show you uh for your next experience unify the navigation what search applications are associated with it and then from this table you can see you know the also the uh Associated search uh configuration and you can see if all the zinc configuration has been migrated to AI search configuration and if not you can start taking action and then migrate use uh based on this form and once everything is already migrated to AI search then you're all good you can now turn on AI search index experience all right I think this is all the content I wanted to share thanks again for watching thanks this is really helpful tool I'm looking forward to hear your feedback on it uh once you installed it again if you still have any more questions you can reach out to us in the community um you also can uh see this recording again it's going to be uploaded to YouTube really really soon and thanks for joining today thanks everyone thank you