foreign Academy if you are interested in additional resources around these topics please do visit the AI and intelligence community and be on the lookout for additional AI academies in the future my name is Gerard dwan and I'll be walking through getting started with external content for AI search I have extensive experience in a search as as a whole for over a decade and I will be walking through specifically how to just get up and started with working with external content for AI search in the servicenow platform before I get started there may be some questions that are that result in looking at some features that are a little bit more forward-looking so I do want to put it our Safe Harbor notice out there in the event that occurs from a high level our goals today are to have a overview of the external content as a topic go through an exercise for a few minutes and provide the opportunity for some question and answer that the reality is that AI search uh the reason why we started focusing on it is because the expectations for uh the workers right employees within organizations have changed pretty drastically there is an expectation um that the same consumer experiences they get site out get outside of the Enterprise They want to get inside of the Enterprise and so that's one of the main drivers behind AI search and why we focus on AI search being this this consumer grade or consumer quality product customers that have used AI search so far have experienced better relevancy especially when it comes to the the Legacy system but in general we've seen this as well an average click position basically how far down the list somebody needs to click to get on the answer that they're looking for is a better than two which it which means that people typically find what they're looking for within the first couple of results of the search results set so that's great we're also seeing that the click-through rate is about 70 so when people land on the page they're finding what they're looking for and on top of that what we didn't realize when we started this journey was we're also seeing that when in cooperation with other platform capabilities there's also an additional benefit so as an example here nlu intents for VA can go down drastically if you back it with AI search so all good things that are coming together as part of the AI search um in the in the platform and with that being said I'm just going to touch on our key pillars for AI search basically when we look at AI search and all the capabilities and features that we're wrapping in as part of the AI search capability we want to as we go through this make it natural smart and easy right so just a quick high level on AI search before we get started for specifically the external content topics for today I want to talk about external content very briefly I'd like to talk about when when we look at external content what our overall design capabilities are and then um on once we go through those two things I would like to talk about um the different ways that you can incorporate external content today so let's get to it uh just a couple of quick definitions here external content the what we are trying to do is Index right information that's outside of your servicenow platform and that could be you know the information that lives within documents that are stored somewhere else as well as the metadata that's associated with it right so for a Word document on a SharePoint site it was updated at a specific time right so we want to make sure that that information is available as well as the content uh that's in that document on top of that AI search will display the external content in line with the other information that you have as part of the cert your servicenow platform so users don't need to perform anything special to get up the external content and they are basically using it in an organic way they don't need to know that it's coming from an external source that is of course until they click on something right um the indexing process doesn't just take information from a third-party source and bring all the information in wholesale we'll we'll take the information that you're asking us to and index it and then provide a reference to that that Source right so as soon as you click on I'll use the SharePoint example again as soon as you click on uh one of your SharePoint results you'll get brought to where that document exists within SharePoint so that's a very quick overview of external content from an architectural perspective what we're really talking about is um without external content of left to right right on the left hand side we have our uh not servicenow data the things that live in tables within the now platform we're going to go ahead and index that information and make it available in the experiences on the right hand side I will caveat that Global search and workspace search are both in Early Access right now but that's not something that you're interested in you can you can also use it there and then on top of all this that Now search index search index is also where we do all of our analytics uh offline as part of the the overall processing now when we bring external content into this we want to make sure that the external content gets the same benefits that the servicenow data does so that's why that external content is placed right there we're going to index that information and make it available to these sources what does that look like when we bring the information in the reason why we take this approach is we want to make sure that it's a high performing system when you submit uh when one of those experiences like the portal submits a query to AI search we don't have to go to all the external systems to try to correlate and bring all those results together all that information is going to come from the search index it provides a unified list of results so that means that everything is relevancy ranked as part of this information as part of a unified information list so you can say you know the the results are mixed based on the relevancy on top of those two we also have machine learning relevancy that is available across all of the information regardless of whether it's internal or external so you can use that external content for understanding what the what is relevant about it and allow the machine learning process to take over it in that part as well on top of these things we also because we're treating it as pretty much the same as everything else that we have in our system we can do all the same configuration from a AI search perspective on that content as well so uh as an overview as a as a example there was a session previously that talked about some of the tuning capabilities and one of those items was result Improvement rules so you can have result Improvement rules that that work on your external content as it would on your internal content as part of your now platform content so that's our our overall look at what the architecture and why we chose this architecture now we have a few different options when it comes to how we integrate third-party sources and they really come in three flavors uh they come as part of available connectors that are in the store now they come as connection capabilities so whether it's a rest API or part of an integration Hub uh bespoke and you also have the option to work with our partners to get even more and more connections to additional sources so I'm going to start with our available connectors those ones are probably the most straightforward we have access to them uh within the servicenow store today the SharePoint connector is part of employee Center Pro um it the caveat with the SharePoint connector is as of today it doesn't include content security but that is targeted for later this year so our guidance for customers is uh our guidance for customers is to use public sites uh for indexing SharePoint content today but know that uh in the not so distant future or later this year uh we should see uh the content security there as well additionally we have the Confluence connector that's available in the servicenow store it's available through a third party this does provide content security as part of it so those are the quick uh service now available connectors that are in the app store today right now whether there is a kind of whether it is the available connectors either of the available connectors or you do it yourself the connection mechanisms are actually the same so you have the ability to use either a rest API that is part of the platform it's an AI search specific rest API or you have the option of the integration Hub that we touched on earlier uh the rest API is really the push mechanism for bringing information into AI search so if you have a client that you're ready that you can keep open and push content um to AI search then the rest API is good for you on top of that if you're familiar with API usage and you have a connection framework already you are absolutely welcome to use this this endpoint the the only thing that is required to make the rest API work is um ensuring that the external content for AI search plugin is is available for integration Hub uh this is really a pull mechanism so this is bringing information into pulling information into AI search I think that it's more intuitive and it's it's very supportable it's a nice editor and I'll show you that in a minute in the exercise the other things that are required are that it requires the external content for AI search plugin integration Hub and then part of the integration Hub being the AI search spoke basically the the framework that is able to speak to AI search the other benefit that's that's not listed here is there are existing Integrations as part of integration Hub that third parties can also ship and provide so we'll be going through a demonstration of what that would look like in a subsequent in the exercise foreign like I said the the DIY options are are the ones that are available and our partners can basically build Integrations um to uh to third parties and push information to AI search so um that we do have a a partner ratio on that has a fair amount of fees outside of the servicenow ecosystem um and they may be available as well as others to build content uh connectors for you so these are all of the options that are available um from that perspective so before we go into the exercise if we could I'd love to submit a poll question and let's see there we go [Music] foreign this question is which external contents would be most valuable for you to integrate with and there are a few options and I'm I'm seeing them come in now I've got good participation Gerard why were you thinking the poll I think we have a question um so I think earlier you mentioned uh something around the nlu intents so can you please clarify what you mean by nlu intent being reduced sure uh we had a few customers that were working on um a virtual agent and providing the best experience possible for their customers and for their end users and what they were doing was really focusing on writing or creating an intent that would help and assist with um categories of questions or or basically knowledge based articles right um they were focused on working on a path from a specific question to a knowledge based article or category of knowledge base articles so what we're able to do from an AI search perspective is use the mechanisms like if you're familiar with them genius results as well as an organic search that really short circuits some of those um some of those intents that need to be written so it's it's a matter of reducing that effort especially for a first pass but ongoing we've found that a lot of customers don't need to focus on creating all of these intents and mapping them to specific articles or pieces of knowledge they can simply rely on AI search to provide the answer or close enough answer for most questions it's a good question great yeah Gerard the the the the benefit for your search enabling it in Virtual agent I think there was a recorded session on the VA Academy where we cover that it does help a lot to reduce the amount of topics that you need to have to be able to go live and still have meaningful content in your virtual agent that's great question so I'll stop the poll let's do it I think you can see the results we have a clear winner here yeah it looks like it looks like folks are uh are seeing that there would be more the most value perhaps in in SharePoint um I'm also I see websites here so I think that would be a good um follow-on because as you saw uh from our um session there isn't a clear cookie cutter um web approach so I think that's that's really good information so thank you for that thank you since we already touched on SharePoint um we have a question do you need to be an admin for SharePoint to integrate to AI search or reader credentials are enough uh that's a good question um so this is a this is a tricky one if you are a SharePoint expert then you likely know uh part of this answer already but you do need um elevated permissions you need read write permissions uh to index from SharePoint and that's because uh the way that my understanding and I'm no expert on pure um SharePoint Integrations but from our perspective the way to get Deltas required to read write access so there would be you know our SharePoint connector basically supports that bulk load or indexing in addition to the incrementals and so in order to get incremental reads um there there is a need to do to have read write permissions so it is elevated it's it's an admin level role yeah great so um because I got the same question multiple times yes this session is recorded and will be uploaded to our community um you can move to your exercise right now thanks great and I should have used that time to make sure I was still logged in so give me just a moment we've got multi-factor and everything involved here um so great questions so far really appreciate the participation what I'd like to do as part of this exercise is twofold and I'll I'll have a talk track here while we while I pull everything up again what I'd like to do is twofold first I want to show uh regardless of the mechanism that you use to integrate whether it's the rest API or it's the integration Hub we are going to be creating an external schema and I'll talk through that in a moment um and then we'll we'll look at the rest API endpoint uh briefly just to get a feel for it and I'd also like to demonstrate uh briefly again uh what that integration Hub interface looks like [Music] and only because I had to move around screens I'd like to just confirm that [Music] we can see this endpoint okay great so we have a couple of things that I'd like to show as I mentioned we are going to take a look at what creating an external schema looks like this is essentially creating that tabular structure that we can use um to to do all the configuration and tuning that we talked about a little bit earlier so if I go to actually I'm going to type an external here in my filter nav and you can see there's an AI search section and under that AI search section under external content there is a section for creating a schema so what this does as I as I mentioned briefly is it it creates a schema for you to interact with basically within the system um I'm going to just call this you test test as a mechanism to show you what this looks like uh once you've created it and it'll basically look like by the way um creating any other table within the servicenow platform right you have the ability to update which columns are available and so if there are different types of metadata that are important for your Source system you can go ahead and insert that in here uh and then once this is created you can interact with it the same way um that you interact with any other table within AI search so uh because of time I'm not going to go through all those processes but you may have seen in a previous AI Academy this ability to do tuning based on specific fields that exist or or column values that exist within a specific search source so this is all part of that process I will also briefly review what the integration Hub looks like so the way that I like to get to the the integration Hub is through the flow designer interface so I'm going to just jump to flow designer and you'll see I I have a pretty full environment of lots of test connections and all that fun stuff but what I will show and some will call it blasphemy and that's that's okay it's just an example what I will show is a very quick configuration of something that uses jira uh so this is a what's called a jira connection it's part of an existing integration Hub spoke um you can see that it's currently configured and there are many other options uh for integration Hub spokes um but I do have an existing flow that I can show you briefly what that looks like in order to demonstrate the capability there so here's my existing connection and then if I go to flows and I do a search for jira you can see that I have an existing flow called AI search index jira content and within this we have a series of steps or actions that are communicating with AI search and the source system and then ultimately sending information to the AI search index for processing so if this is for example there the series of steps includes initializing a bachelor a bachelor is basically a message in um step that basically says hey we want to send a certain number of items to the AI search index so that's what this spoke action is doing and then there's a nice spoke action as part of Shira that's basically just saying get me the issues and then we're using a mechanism for each so for each of these we want to ingest the document right the AI search spoke action is ingest document which basically pulls information from each of the issues that we're receiving and then assuming that I'm going to just say that assuming the bachelor is full uh so that's the there's a threshold in The Bachelor for how many documents we can process at a time uh we'll commit the information that we have and then ingest the document so regardless of whether we've committed and then we'll ultimately commit the index and release the Batcher to clean up so during this process we'll take all the issues that match this criteria the jql for Ajira and then pass the information ultimately to get index released into the bachelor if we did everything right uh what we'll see as part of our experience on the portal is a mixed list of information that is available uh to the end user so what that would look like is [Music] if I do a search but my my jira account is just sample information so it is just information about llamas we're big on llamas that servicenow so this is basically my my list of of llamas here as part of that jira cert Source um so you can see that that these have come through successfully um but otherwise I would get a typical search result um as part of searching here so this is a quick uh end-to-end on configuring an external content source and then ultimately seeing results as part of that process so with that in mind I am going to see what kind of questions we have [Music] uh there's a lot of good questions uh Nabil I want to do a time check um we're good we have extra few minutes five minutes maybe we can go over these questions so in the order that you prefer great so uh let's see uh first there's a question about um the media recording what the meeting recording will be made available so thank you for asking that is AI search limited to uh ESC portal or and or all custom or existing custom portals so um the AI search is available for portals based on service portal um we do have a couple of customers that have highly customized their service portal or ESC and it does work fairly well but it needs to be based on um that service portal um technology or framework uh we do have a verbal question I don't know if we have time for that [Music] um other than sorry no go ahead we still have uh other than facets how can we highlight the content Source being external uh great question there's a couple of quick things I'm actually gonna just uh so I do have um some SharePoint content in this as well um so I did want to share that from each of these can have their own rendering card um they also get not only a facet here but they get put on to the the navigation tabs as well um so those are a couple of different options um that are available um does this always need to be a document or can it be a URL and short description uh the type of metadata that or the type of information that gets passed it doesn't need to be a document we consider all AI search results to be a document of some kind so we actually have instances uh within the now platform there's a something called external links and that's actually a really good type of search result because it's just a URL a title and sometimes a short description so that's something that we can provide as well um as long as it fits roughly that framework foreign [Music] so our question is regarding the preference for building these um external connectors um whether you prefer to use an existing one build it yourself or get it from a partner uh back to you great thank you see in the process are you consuming external data via API is that the get step in the process I think this was relating to the the flow designer step that I that I got here um the get issues is basically the part that is speaking to the the third party system and then grabbing the information and bringing it back so I believe that's you have it right yeah uh there is a question that's out okay can you please show for a custom connection like a feed from alert any learning portal uh I I'm not certain on that one um that's something I'd have to follow up I'm not sure I'm not sure the context of that question foreign you can leave it in the community and add more context to it we'll make sure we answer it great awesome that's good qualification so um I think we can end the poll we're we're a little over um but it looks like most folks would use an existing connector and then there's some confidence in in building it um building it on your own which is great yeah definitely Gerard thanks a lot for today's content as I mentioned this meeting is recorded will be uploaded to our YouTube channel and posted in the community if you have any questions please feel free to leave them in the community and uh looking forward for the next session thanks everyone [Music] thank you