Activating AI for
SAM and HAM

Christine Morris

Welcome everyone. Today we are going to be talking about activating AI for SAM and HAM. It's one of our most exciting use cases in the AI landscape these days. So we're going to spend some time taking you through that. We're going to do a little welcome and introductions. We're going to talk about Pro Plus and how you can unlock those Now Assist AI features using AI agents in your asset workflows. And then just an introduction around general AI on the ServiceNow platform. And then talk about — are you ready, and what do you need to do to get there. We'll save some time for questions.

Before we dive in, I want to talk a little bit about who Ondaro is. If it's your first opportunity to attend one of our MasterClass sessions — we are the only pure play ServiceNow partner with fully certified resources across the platform. And I do mean everything. So we've got expertise in ITSM, ITAM, CRM, SPM, App Engine. One of the things that really sets us apart is that typically former practitioners — folks that have been doing this and working in their respective fields for many years — lead these practices. So we're very much specialized there. And we really do work in these three pillars. So envision — if you're thinking about a transformation for your business, if you need organizational change management support, we're doing a lot of AI readiness assessments because everyone is so excited about AI, but you've got to make sure you have your best practices and your data in place for that.

Our implementation practice covers platform architecture and engineering. We have award-winning UI/UX designers. We've actually got two portals in the top ten for this year's award, one of them in the top five — potentially going to be number one. We're pretty excited about that. Application modernization — we do a lot in the App Engine space — and then manage and optimize. We know that practitioners on the platform don't always have all the resources they need. So we've got steady state operational support that we do as well.

I am Christine Morris. I am a senior director here of consulting services. I lead all of the various practice areas in the ServiceNow ecosystem — and I think I've passed my decade mark. I spent about half of that as a customer, sitting in your seats myself, but pleased to be here today. And I'll let Ian introduce himself.

 

Ian Cahall

Sure. Yeah. Hi. Ian Cahall. I'm associate director and principal architect for Ondaro's IT practice. I work in a variety of areas, but over the last several years have largely been focused on asset management — HAM, SAM, enterprise asset — you name it. I've been working in and around the ServiceNow space for a little over a decade as well, prior to getting focused on ServiceNow. I was doing IT operations management for Fortune 500 companies. Also happy to be here and happy to be talking about this topic with everybody today.

 

Christine Morris

Just a little housekeeping. We want this session to be interactive. If you've got a question, post it in the chat. If you need help, post that in the chat as well. If you need captions, click on the little more dots at the bottom — you can get to captions and turn those on.

If you're new to our MasterClass series, we've got quite a few. We've done the CMDB MasterClass with about ten sessions that you can access — lots of great, meaningful information on how you manage your CMDB and that whole lifecycle. This is our sixth session for ITAM. We've done other sessions on HAM and SAM, basic asset lifecycle best practices. And then our most recent MasterClass is our Platform Owner MasterClass, which we've just had our first session on. It's been highly productive with lots of great engaging information. My assistant will post the links to those in the chat. So if you're interested, please check them out.

And if you've been to one of our MasterClass sessions before, you know that we are famous for our polls. So just a quick poll for you. What's your current relationship with AI — and not necessarily specifically on the platform; we know a lot of people are just getting into that — put your answers in the chat. A: ChatGPT or your AI assistant of choice is your best friend — you use it for everything. B: you use it for work. C: no, I'm scared of it. D: you've been living under a rock — AI, what is that?

Oh, we've got lots of A's — I love it. I'm an A as well. A couple of folks are just using it for work. It also does great things at home — like grocery lists and meal planning. I use it for everything. No one is scared.

 

Ian Cahall

That's good.

 

Christine Morris

At least not yet, right? All right. With that, I'm going to turn it over to Ian to start diving in. Ian.

 

Ian Cahall

Yeah, thanks so much. So we are going to get started talking about using Now Assist for asset capabilities. What I do want to note is that as part of this conversation we're going to focus on a few things that are very much focused on current state. Consider everything that we're talking about as valid up to Zurich. As many of you likely know, there have been some recently announced changes to the platform, and we will talk about that later in the session — how that changes some of this — but we wanted to make sure that we stayed focused on everything current up through the Zurich update, because for those of you looking to do this right after this session, you'll need to know how it works up to this point, not so much the future. Be aware — we will talk about that current state, but we will also make time to talk about how things might change and what that might look like in the Australia update and beyond.

First, let's talk a little bit about the basics. What is Now Assist and how does it work? Now Assist is ServiceNow's AI model. Unlike a lot of the AI models that you see out in the world — you think about ChatGPT, which has GPT — I think we're on 5 now. Of course Anthropic has Claude Opus and now the very mysterious Claude Mythos, which I'm sure we're going to learn a little bit about in the near future. But for ServiceNow specifically, Now Assist is what operates the AI platform.

And when we think about AI, it seems like most of the folks here are familiar with the fundamentals of AI. We're thinking about a few key tiers of AI today. That's predictive AI — which is using an LLM to analyze data and provide outputs. There's generative AI, which uses prompts to derive text or other content — could be images, could be video. There are a lot of those sorts of things coming out of generative AI. And then of course agentic AI, which is where we're giving a system of LLMs a goal and asking it to output a complex action. And we'll talk about all three of these, plus a couple of new layers, as we get deeper into the session.

When we think about the generative and predictive AI capabilities of Now Assist, oftentimes we are doing that via a prompt. And this kind of exists in a few different spaces within the platform. For those of you that are familiar with Now Assist and have used it before, you may be familiar with the ability to chat directly with it via the Now Assist panel, but there are also other systematic prompts that get injected into Now Assist when we prompt it via UI actions or other things.

 

And so oftentimes this is going to look something like what you see on the screen — something like "summarize the software asset estate's overall compliance position." Now Assist is going to take that information. It's going to look at the scope with which it's been granted access. It's going to give you an output. And this is going to look different depending on the data it's working with. You might receive something like what we see here in this example — you've got 250 publishers under management, 23% are fully compliant, etc. At a surface level, this all makes sense. For those of you who — based on our survey results — are either using some form of AI for work today, or are very much an AI enthusiast — these things are going to be pretty straightforward. But we want to get more into deriving value from Now Assist. That kind of output is very rudimentary and very straightforward. The goal with any kind of AI implementation within an enterprise environment is going to be to produce actual business value.

And so as we take a step forward, we want to talk a little bit about how to get Now Assist. And again, this is very much focused on current state up to Zurich. If you are licensed to do this — and if not, you do have the ability to buy it — you can request Now Assist for Hardware Asset Management or Now Assist for Software Asset Management on the ServiceNow Store. There is a Buy button as you guys can see there. If you are currently licensed for this, it's as simple as going out to the Store — that button should reflect Get instead of Buy. If you are not currently licensed, you can select Buy and go through the process. Obviously, for larger enterprise customers, you will likely be encouraged to have a conversation with your ServiceNow account rep to go through those conversations.

Once installed, the process is super easy. You can go into your plugin manager on your platform. We'd recommend starting in a non-prod instance, finding that plugin via the application manager search — in this case, Now Assist for Hardware Asset Management — and actually installing the plugin. Once that plugin is installed, you do have to take a couple of quick additional steps. Number one: make sure that we enable the Now Assist panel, if it's not already done. That can be done by going to your All menu, going out to Now Assist Admin, and then Now Assist Experience, and actually toggling on that Now Assist panel. But this is just the first step. This is how we get Now Assist onto our platform. This is not how we actually enable Now Assist to make it usable for our asset use cases.

That's where we want to go next — and that starts with AI skills. For those of you that aren't familiar with the concept of AI skilling: an AI skill is something that provides additional context to your AI model and gives it additional instruction, and structures prompts to allow it to be used for the specific use cases we're focused on. Now, as we can tell from the slide on the screen, SAM has received a little bit more love than HAM when it comes to the currently available Now Assist skills, but there's definitely a lot of value in both places. Both HAM and SAM do have pre-built AI agents — and of course we'll talk a little bit more about agents in a moment. But there are generative AI skills that exist for both.

For HAM, we have Generate Hardware Asset Insights. This is something that's accessible via the hardware asset workspace. And then for SAM we've got a few different things that I think present a ton of value. We're going to talk a little bit more about these in a moment. But we do have our basic summarization. So we've got publisher and product compliance summarization. This is going to help you quickly sift through your details as far as your current software asset compliance position. And it makes things a lot simpler than — for those of you already using SAM today — digging through your workspace, looking publisher by publisher or product by product, trying to understand where and when you need to take action to improve your compliance.

In addition to that, there are skills for recommended actions. Those of you that have used both HAM and SAM will know that there are important actions that get surfaced already as part of the workspaces for each. But this actually goes a step further. When we see the important actions in the hardware or software asset workspace, they're very surface level — it's going to tell you that a certain value is missing or a certain field needs to be updated, it's out of date, etc. These recommended actions go several steps deeper — the best things that you can do both on and off the platform to improve your current software asset position, whether that's optimizations you can change, contracts that maybe need to be updated, etc. It uses the AI capabilities that come with the skill to make those important actions become much more actionable.

The SaaS User Resolution skill improves the ability to consolidate SaaS users across platforms. For those of you that have used Software Asset Management, you know that a fairly frequent issue is that across an organization you may have different naming conventions for your software users — some may reflect your company's domain name, could reflect a specific string of letters and numbers, etc. This SaaS User Resolution skill uses AI to consolidate those users into one record, which is great for a number of reasons. Number one, it makes things a lot easier to manage. But number two, it also improves your costs when it comes to your software asset management licensing. And then finally — and this is a very new one — the ability to extract software entitlements from a contract. We're going to spend a little bit more time looking specifically at that one in a slide. This one is hugely valuable — maybe in and of itself a great use case for why you want to adopt Now Assist for Software Asset Management.

 

In order to do so we do have to activate those AI skills. It's important to note that when we actually install the Now Assist plugin for HAM or for SAM, some of these skills will be activated by default. But it's always important to double-check, because not all of them will. When we activate that plugin, the next step is going to be to make sure that we activate these specific features. We'll go into the All menu, open the Now Assist Admin console, and open Features. We'll find those skills we want to activate. They'll be represented via feature cards. We'll select them and click Activate Skill. Also important to note that in some cases, some of these skills do have certain behaviors that you'll want to configure, like where we want to specify where we see the outputs from the skill. For a summarization skill, you do have options. You can either have that be presented within the product — so if you're in the workspace and you click Summarize, you'll get a blurb that pops up there in the workspace. Or alternatively, maybe you want it to pop up in the Now Assist panel so that you can continue a conversation with Now Assist. You do have that ability as well.

Also note that for installing those plugins, you do need to be a platform admin. And for activating AI skills, you need to be either a platform admin or a Now Assist admin from a role perspective. Make sure you have the appropriate role and access when attempting this. So when we think about some of the things we've just mentioned — with the summarization capabilities and the recommended actions — you do get a lot more value out of your asset workspace.

Some of the things that we've already talked about, but I think are worth mentioning — and I will jump to a better screen that'll make this a little more legible. Some of the things that we can do with that summarization and those recommended actions are prioritizing what is most important for us to do on a daily basis. For those of you working in Software Asset Management for your day-to-day job — you may have yourself looking at the software asset workspace two or three times a day, spending 15, 20, 30 minutes each time trying to get to the root of what you need to be doing. These AI skills actually: A, reduce the amount of time you need to spend doing that, because they're going to surface those trends and highlight those priority actions. But B, they're going to make sure that you're not wasting time on things that aren't important or aren't critical — like re-looking at things you've already actioned recently, etc. And by doing so, you're going to save both time and money. Because number one, one of our core goals in doing asset management is optimizing our asset spend. But number two, the labor we're spending on doing that is also something where we can focus our time more effectively and get more out of both our investment in the product and the practice at large.

 

And hopefully this is a little bit more legible than the previous screen — you can see here in a couple of different places the recommended actions on the right side of the screen. These are recommended actions generated by Now Assist. Some of those things are going to be — as we can see in these examples — health check issues. So specific things that have come up in your SAM health check that require action, and normalization suggestions — specific records within normalization that require some kind of update. And the alternative to this, if you've worked within asset management normalization, is you kind of have to find those things out on your own. You may get a table with a bunch of suggestions and you're not sure where the effort needs to be spent. This helps you focus that in quite a bit. And then in the middle of the screen, we can actually see our compliance summary for SQL Server.

This is an example, but the idea here is all of the data that we would previously be able to see by clicking into specific records, opening up our entitlements, opening up our optimization suggestions — all of this information gets synthesized and surfaced at the top so that we can see specifically where we need to take action, rather than trying to figure that out for ourselves. This is super helpful, especially in more complex products like SQL Server. Some of the context you're going to get with this Now Assist skill is context that maybe you don't have, because you're not an expert at how SQL Server is licensed. Obviously every environment is different — and we'll get into why that's important in a little bit. But generally Now Assist has that context that comes with expert-level experience managing software records across a number of different industries and verticals. So the idea here really is — how can we take what used to take half an hour, an hour, several hours even, to sift through our current SQL Server position, and turn that into a couple of minutes of reading this summary and then knowing where to go from there?

And then finally — I mentioned this previously, but I think that this is something a lot of organizations will find value in. One of the biggest levels of effort when standing up Software Asset Management, and also one of the largest time consumers when it comes to maintaining Software Asset Management, is the upkeep and upload and import of software entitlements. What this capability allows us to do is — where previously we had to take your software entitlement information and translate it into some form of a spreadsheet that could be imported into ServiceNow, and then work through that to resolve any errors or conflicts — this allows us to take your raw software contracts, purchase orders, and they can be in a number of formats that we previously wouldn't be able to use — like PDF, or image formats like PNG and JPEG — and we can actually use AI to read those documents and create entitlements from them.

 

Now, it's not completely hands-off. You do still have to review the results of the AI import — much like you would want to for anything that you've got AI working on — just to make sure there's a pulse check that things are fair and valid. But you can review those things. If there are errors, you can certainly address those and approve the entitlement record creation. Now, this is something that historically likely took anywhere from 30 minutes for a single contract up to an hour — maybe even two hours — depending on how many software products you're purchasing within that contract, and distilling this down into a workflow that takes a couple of minutes to complete.

And I'll zoom in a little bit so this is easier to read. For those of you that are familiar with how Software Asset Management entitlements work — the really great thing that we're seeing here is this process, which is quite literally three steps. You upload your contract document — and again, it could be in any of those formats. Now Assist will extract the entitlements. From there we review the entitlements just to make sure we're happy with the outcomes. And the really critical thing you'll note is that it's recognizing things like publisher part number, which drives a lot of our decision making around creating these entitlements, as well as things like your metric group and your product category. This takes a lot of that effort and distills it down into a couple of minutes rather than a much longer process historically.

 

Christine Morris

I am going to do the next poll. And Ian may not realize that the bugs in the computer have taken him off video, so maybe we can get his smiling face back on. Which AI capability sounds most valuable to you? A: automatic data summarization. B: data qualification and cleanup. C: the pre-built AI agents. D: custom AI workflows. Or E: give me everything — I want it all.

Already getting votes for data qualification and cleanup. I knew it — when you see that cost saving amount like that, that's a game changer. Lots of B's. Everybody's interested in the data qualification and cleanup. I know that's one of the biggest pain points. One E: all of the above. And the custom AI workflows — love it.

 

Ian Cahall

So let's talk a little bit more about that. Next up we're going to spend some time talking about AI agents for asset workflows. Everything that we talked about previously — this is your out-of-the-box Now Assist AI skills. You just plug and play these. You install your plugins, you activate your skills, and you get to work using them. Super simple to set up. And we will talk a little bit about some prerequisites in a moment. I don't want to understate those. But again, the idea here is that this is something that — in terms of turning these things on and starting to use them — does not require you to be an AI guru by any means.

As we get into AI agents, this is where things get a little bit more complex. We'll talk a little bit more about those. First, though, I want to talk about what is an agent. And again, for most of the folks here, you're on LinkedIn, you're hearing and learning about agents. This concept came about early last year. It had existed prior to that, but it really kind of hit public common knowledge last year. The idea here is that an agent — knowing all of those things we learned about predictive and generative AI — is something that has the ability to take those capabilities and translate them into specific outcomes when given a goal, and do that largely unguided. It can make certain decisions on its own that drive more complex outcomes than just summarizing some data or extracting data from an image. As you can imagine, the ability to automate your processes and drive more value for your organization with an agent is much higher. Specifically on ServiceNow, this is how an agent works.

First, we give the agent a goal. In this case, just for an example, we're going to say we want this agent to handle software requests. The AI agent is made up of that LLM — Now Assist in this case — the memory or the context that we give it, which exists across your platform. Specifically for ServiceNow, we're going to focus those agents on specific data sets. And today we're focused on asset management — so that's where we're going to focus our agents. And then we give it tools — the ability to do things like call workflows, call integrations, and take specific actions. And then we look for it to complete those actions. In this case it can take a user request for Microsoft 365. It can recognize that Microsoft 365 is a certified software within your organization. It can allocate an available E3 subscription and check to make sure that those allocations are available — and can even kick off provisioning and installation, assuming that we've given both ServiceNow and the agent the capability to touch those things. This is just one example of what an agent can do on ServiceNow.

And the really important thing that I would stress when we think about agents is, generally speaking, there's this notion of one agent equals one thing. And oftentimes when we are trying to get really complex outcomes, we look for those things to be done via a variety of multiple agent groups or pods, if you will. And the only way we can really do that is understanding specifically what our inputs and outputs are. When we look at our available agent workflows that are there out of the box, there are a few specific ones for both Hardware and Software Asset Management. On the hardware side, we've got the ability to help manage those hardware request workflows, but we can also help with hardware asset repair workflows as well. On the software asset side, we've got the ability to help manage software request workflows — just like our example showed earlier. But we've also got the ability to evaluate software removal candidates. And as you can imagine, that's something that provides a ton of value — for a lot of folks, you're spending a lot of your personal time and energy looking at those removal candidates and trying to work through them.

I want to highlight really quickly what this looks like in specific action with the hardware asset repair workflow. As I said before, in order to get some of these outcomes, we really do need a team of multiple agents to get from point A to point B. What this looks like here is three specific agents. We've got the Repair Asset AI agent, which has the ability to perform web searches and provide details for repair steps for a technician to follow. We've got an Evaluate Asset AI agent that's got the ability to also perform web searches and provide detailed evaluation guides for technicians to follow. And then we've got a Next Best Action AI agent. This is basically the orchestrator — making sure that we know which steps to take based on the inputs and outputs we get.

A simple example: your asset manager within your organization creates a repair order for a laptop. Your Next Best Action agent is going to establish which tasks need to be completed and validate the task detail. So it's going to make sure we've got information associated with the asset record — like warranty, like manufacturer — those sorts of things we need to know to effectively solve the issue. It's then going to pass that off to our Evaluate Asset agent. That agent is going to find a guide on the internet to help our technician evaluate whether or not we can repair the device, or if it even needs to be repaired. The technician will follow those steps, confirm the troubleshooting result within the ticket and within the conversation with the Evaluate Asset agent, and depending on the outcome — whether it's repairable, unrepairable, can be redeployed, or is under warranty — that issue can be diagnosed and will then call the Repair agent. The Repair agent is going to find the specific steps to solve the issue that the Evaluate agent determined and provide that to the technician. From there, it's going to continue to support the technician as they work to fix the device. If the first set of suggestions work — great, we can close the task and close the ticket. But if not, it'll continue to help until it's reached a point where it can no longer solve the issue — at which point we will either decide A: we have to dispose of the asset, or B: maybe we need to set up an RMA if we can do that.

The idea here is, number one, you're saving a lot of time and not being reliant on tribal knowledge. But number two, you're also saving a fair amount of money on unnecessary repairs outside of the organization where you have to pay a third party to complete that work. And of course, this is something that anybody within your organization has the ability to work with, assuming that they are a fulfillment technician working within that task flow.

Some other examples of agent capabilities on the platform — beyond those out-of-the-box workflows — are your Agent Studio, and there are a number of out-of-the-box agents that you have the ability to access, with no major configurations beyond where those tasks need to get assigned. But you also have the ability to customize and create new agents for specific purposes, depending on what you're trying to achieve. Those customizable agents do need to be given a context and a data set to work from, just like we talked about. Sometimes you need something to happen that goes across different modules of the platform, and this is where a custom agent may be more appropriate. You can kind of compare this with a decision table in your Flow Designer. The same concept applies — ultimately we're trying to automate the decision-making process. But here we're given a lot more opportunity to leave some of those variables open-ended, so that context can be the decider around how we treat certain things and why. The same is true for Flow Designer. Flow Designer does have the ability to make out-of-the-box and custom AI agents callable from the flow. This obviously gives us a lot of ability to extend to our spokes, our Service Graph Connectors, action packs, etc.

One example: perhaps you want to create a workflow that calls an import from SCCM, evaluates changes to that data set, and then uses that information to take a specific action — maybe retiring records or updating states, those sorts of things. This gives you the ability to do that in a much more flexible manner than you used to be able to with very rigid inputs and outputs. And then finally, the ability to take off-platform action. As we've talked about, your agent outcomes are enhanced when you've got multiple agents working together — that's true both on the platform and off the platform.

A lot of you have likely heard about technology like MCP, or Model Context Protocol. ServiceNow does have an out-of-the-box Model Context Protocol server technology that you can request from the ServiceNow Store. This grants the ability for ServiceNow agents to work in concert with off-platform agents. So maybe you've got ServiceNow and a third-party technology that introduces an agent, and you want your ServiceNow agents to hand off information and actions to that agent. This is a great opportunity to further automate — where historically you may have had to build a very explicit, rigid integration or had a specific person swivel-seating information from one platform to the next. The opportunities here to open up — number one, not just automation, but number two — flexibility and the ability to accomplish more without having all the foresight you might otherwise need — is really significant when we introduce agents to the mix.

 

Christine Morris

Another poll question for you guys. What's your biggest blocker in adopting AI and ITAM? A: data quality — we know we all struggle with that. B: lack of internal expertise. C: the cost of additional licensing, or unclear ROI. D: governance and risk concerns — I know that not all of our security departments have gotten on board. And then E: all of the above — and my own internalized self-doubt as an adult caused by my parents.

Oh, we got a D. Let's see. Lots of D's — and parents. Some folks — cost and unclear ROI. And Jason — C. Our sales folks can help with that.

 

Ian Cahall

This is a great segue because we are going to get into that "are you ready" segment in just a moment. But before we get there — a lot of what we talked about today is true up through Zurich. The fundamental capabilities will remain post-Zurich. But some of how you consume these things are definitely going to change, as ServiceNow rolls out this new notion of being an AI platform.

Historically, we looked at ServiceNow as a lot of things — ITSM, ITOM, all of these different capabilities. But very recently — within the last couple of weeks — they've really planted their flag and said, "We are all of those things for sure, but now more than ever, we are an AI platform." And here's what that looks like. Previously, there were different tiers of licensing — your different products — and then everything we've talked about as far as activating Now Assist and activating agents was all layered on top of that. It was separate licensing, separate products that you had to activate. Going forward with this new model, AI is going to be included on some level in every product that you work with on ServiceNow. What this means is that many of those things we talked about — activating plugins and activating skills — some of that will still need to be configured, without a doubt, but much of that will be available out of the box with no additional plugin configuration required, beyond your fundamental platform readiness — which again we'll talk about in a moment.

So with your Foundational level, you're going to get those out-of-the-box skills and AI agents that we just talked about. The idea is that AI is going to help your work. It's not going to complete all of it for you. It's not going to be the fundamental decider on how your stuff gets done. Ultimately, your humans in the loop will be the ones driving your outcomes, but they will be assisted by AI. At the Advanced level, you're going to have your agentic workflows, and those workflows are going to help you complete more of that work — just like we talked about in our example for your hardware asset repair flow. Understanding the context, applying that domain knowledge, and automating that front-to-back process throughout the data and the things you're connected to on your platform today. And then at the Prime level — everything that we just talked about, plus the ability to create custom AI specialists, agents, skills, etc. This is where we really get into that full autonomous workload, where there isn't any expectation that a human has to be in the loop to complete those things.

Now, we've got a lot of D responses from the previous question. From a governance perspective, you still want some level of oversight — making sure that people are aware of what's going on and why and how. But the idea here is that we get to a much more autonomous stage. And with that being said, it's really important that you understand how that changes some of the products you're used to working with today. If you're a HAM Pro or SAM Pro shop today, in the future you may become a HAM Advanced or SAM Advanced shop. Now obviously that change isn't going to be a hard cut-over. If you want more information about how and when those changes may happen, be sure to speak with your ServiceNow representative. And then of course, you've got your SAM Enterprise moving to SAM Prime.

Apologies — my camera overheated again. The idea here is — if this is something you guys are looking to adopt — everything we talked about in those first two segments of the session is true today, and true for your HAM Pro, SAM Pro, and SAM Enterprise segments. In the future though, that could change — and it could become something where you're in these HAM Advanced, SAM Advanced, etc. segments and you want to make use of these things without having to go as far out of your way to do so.

Next up, we want to talk about your autonomous workforce. Autonomous workforce is a new capability coming to the platform, in tandem with all these other things we've just spent some time talking about. Previously, we spent a couple of minutes talking about AI agents — how those are set up to do a specific task, but they do require some orchestration and technical setup. This autonomous workforce capability is an evolution of that. Here you're going to get access to specific specialists that do a job — not a task, but what we would analogize to our own jobs. They execute these jobs autonomously as a group of agents — that's really fundamentally how these AI specialists work. And as opposed to the setup we talked about previously — going out and activating plugins, setting up skills, doing some configuration — this is really a matter of selecting which AI specialist you want and onboarding it with a few clicks. What specialist do you want? What group do you want it to be a part of? And when and how do you want it to work in terms of scheduling?

 

One of the great examples of this is the autonomous SAM analyst, which is coming in the second half of the year. There are a number of others that will be available, but since we're focused on IT Asset Management, your autonomous SAM analyst is going to be basically an analog to a SAM analyst that you might go out and hire today. This is a virtual worker that tracks your software inventory, manages your license compliance, automates those access requests, and optimizes software spend without any specific direct human intervention. Now, you do have the ability to customize that to whatever degree you need to — if you want to have people in the loop helping make some of those decisions. But the idea is that this SAM analyst will work completely autonomously around the clock, doing a lot of the things we just talked about — whether it's pulling together data, executing on removal candidates, improving your optimization rules for software reclamations, granting access to software based on requests and availability of licenses. All of these things it will be handling in the background so that you can deal with the more important tasks — like contract negotiations or product migrations for your employees — things that are much higher value and much more important to your organization will remain in the hands of the humans.

With all that being said, it's really important to take a step back and think about how you become ready to use AI. And before we get into the specifics of that, I want to qualify why you need to get ready in the first place. A lot of people ask — well, I can go out and use ChatGPT, I can use Copilot, I can use Claude today, and it's great. Just like we talked about earlier in the poll — a lot of you are doing that already today. So why is that so easy, but I keep being told that I need to spend this time and energy to get ready to use something like Now Assist? That's a very fair question to ask.

The differences are pretty straightforward. When you think about your chat interface AI, your context is constrained to you — plus or minus some training data that's out there, but usually that's the internet at large. And so when we think about what that means for its ability to respond to you, you're going to get the types of answers you're looking for. And many of those answers could even be right, but they're going to be more general in nature — specific to technology, specific to finance, or certain other areas. But they're what we would call general knowledge. As a result, those results are rapid — but their trustworthiness is variable. We all know that. We've all put a question into ChatGPT and gotten a nonsense answer.

And those things are getting better with time. But when we think about what we need enterprise AI to do, we need it to be knowledgeable about our specific work environments. And that information is not public. So very often — if you've got an employee within your organization who needs a new laptop — what is the process to request a laptop within your organization? Who's responsible for that? How do you go through purchasing? An AI model trained on public consumer data cannot inherently know these things — and this is why it's important to prepare and get ready to adopt AI on a platform like ServiceNow. And this is why there is more effort that goes into it than what you might do when asking more general questions to a public chat interface.

So how do we get there? When we think about our AI data foundation, specifically on your ServiceNow platform — and since we're focused on asset today, we're going to focus on the things you need to do to drive that asset data within your enterprise. We're looking at a few specific categories. Your user data, of course, because we need to know more about the people using the technology so that we can provide the right context to them based on their role, where they stand in the hierarchy, their group membership, that sort of thing. We also obviously need to know about our asset data — lifecycle dates, warranties, renewals — as much of that information as we can source confidently. CI data — and as those of you who have attended any of our previous sessions know, asset and CI data go hand in hand. So it's really important that we are confident in both of those data types. Operational info, change logs, relationships between CIs, etc.

Knowledge — and knowledge is a common point you're going to hear regardless of whether we're talking about asset, HR, wherever you're at on the platform. Knowledge is going to be important for sure, because that's going to bring recent process context to light. And then of course, any activity data we can provide — service history, usage on our assets, the people using them, etc. — because the context for this information doesn't exist outside of your enterprise. We need to make sure that what's on the platform and what we're handing AI to work with is valid, and that we're confident in it so that the outputs it gives us — we can also carry forward that confidence with.

And one thing about the way that I position this — I think it's a really good way to think about it. Consider how you would go about answering a question if somebody within your organization asked you — if you had to cite your sources when you answer that question. AI needs access to those same sources to answer the same questions. If you had to provide a full explanation of how you do a specific process, think about that when you think about how AI needs to be able to access data within your organization.

So how do we get from where you're at today to potentially adopting AI? Number one, you want to establish your goals. Consider the measurable criteria that you're going to use to measure success. And we'll talk about that a little more in just a second. Consider a timeline that meets your business needs. Is this something that needs to be done in a month? In six months? A year? And which specific outcomes do you want or need? We've been talking about asset management — so are we trying to get more efficiency out of our asset requests? Are we trying to save more money on our asset spend? What specifically are we trying to get, rather than just "we want AI." From there, understand your specific requirements. Who needs to participate? Who's got to approve? Which specific technologies are involved? And what is our budget roughly?

Think about your approvals. We talked about security and compliance already. But obviously finances, technology stakeholders — all of these different groups have to come to the table to make this a success. You want to assess your data. What is your data quality? Really importantly, what is your data coverage? And I think this is something that a lot of folks overlook. Where are your gaps? What doesn't your data currently cover that you may want it to for your AI pilot to be a success? And then of course, think about things like upkeep and exceptions — because those things will invariably come up should that pilot exit the pilot stage and become a successful AI adoption. From there, you do want to run that pilot.

Within that pilot, you need to give yourself room to prove your use cases. Really important that you test your consumption levels — because AI is on a consumption meter, whether on the ServiceNow platform or off the Now Platform. How much people use it drives your costs directly. You need to understand what it's going to cost you to run and maintain these AI tools going forward. And then from there, assuming your pilot meets your use cases and you've hit your minimum threshold for success, you move to your actual adoption roadmap. And that's where we get into your adopting governance segment. Number one, obviously, we're going to execute that roadmap. But you need to establish an AI governance program — and this is true regardless of how big or small your AI use cases are, for a lot of reasons. Number one, just like we've already talked about, you need to control those costs. But you also need to control your risks. What is using this AI exposing your organization to? Are we open to that? How will that shift and change over time? Where are we going to grow our adoption? How are we going to grow it where it's appropriate? And what does that mean for our organization in terms of resourcing — both labor and technology costs, etc.?

 

Christine Morris

And Ian, wouldn't it also be accurate to say — so if you have customized things, and you've created custom fields and you're not putting things in the fields the system would recognize — that adds an additional layer of complexity?

 

Ian Cahall

It does — 100%. And the great news is you can ultimately train Now Assist on custom fields, but if you're trying to get to those immediate wins — those short-term wins — you need to avoid those custom fields as much as possible. That just continues the already-standing recommendations from ServiceNow and from us, of course, to limit customization as much as possible.

And when it comes to measuring success, there are four key governance KPIs that I would focus on. Number one, your ROI. And I think this is likely towards the top of the list for how everyone is going to measure success — what is your organization getting out of AI? We're going to measure the cost in and the value out. And those things can look different. Number one, there's consumption licensing, etc. But on the value end there are efficiencies that you get across the board — whether that's labor efficiencies, process efficiencies, etc. But to get there you do need strict measurement of your AI token costs and usage for sure.

Adoption — who is using AI and how? Who's been granted access? Who's actually transacting with the AI? And you need good telemetry to get there. This is something that can be enhanced over time with a more mature vision of what good adoption is. Just because somebody uses an AI technology doesn't mean they're getting value out of it. So going from the broad scope of who's using it down to who's using it better.

Quality is super important. How often does a human need to correct AI? How often is it getting things wrong? Duplicate transactions, prompt overlap, AI-driven incidents — all need to be tracked and measured with strong root cause analysis. And this is how you know the organization is really mature with this — you've got that quality measure of the outcomes coming from AI.

And then finally, of course, risk. What is our AI exposing our organization to? There are a couple of different things we're going to look at — AI model risk profiles, as well as controlled data access. What can AI access within our environment, and are we okay with that? Those things drive a lot of different approvals for getting your AI deployed in the first place. For a lot of organizations, it requires a really strong existing risk management posture. But it is critical for those of you that operate in compliance-heavy verticals.

Obviously, we focused a lot on ServiceNow's Now Assist capabilities. But it's important that you understand that the direction ServiceNow is heading — you're not locked in to Now Assist. Today, Now Assist is where ServiceNow is bringing you these capabilities out of the box. But if you've got other things that you've built — maybe you've toyed around with something custom, like Claude or something like that — you do have the ability to bring that into ServiceNow to drive that decision layer and your action layer, even within your environment. And that's enabled by those workflows we talked about before. You can bring in data, you can drive these workflows, you can have AI make decisions for you, and then you can port those actions out via your workflows.

And so as we get ready to wrap things up — questions to ask yourself: What is your current confidence level in your underlying CI and asset data quality? What compliance requirements do you need to meet to use AI? Which specific AI use cases do you want to take advantage of? Do we have the right operational oversight to govern and control AI use? How can we measure the impact AI will make on our asset workstreams? And do we have the right executive sponsorship for our AI use cases? If you can answer all of these questions with a high level of confidence — for the yes or no questions, the answers are yes, and for the what or which questions, we've got a pretty strong answer — we're confident in what's there. Then you are likely ready to dip your toes in the water and start using some of this technology. And if you're not, there are a lot of options for us to provide you some help. If you're not quite sure how to answer some of those questions, or there are things you need a little more guidance on — if you'd like help developing your AI roadmap, activating some agents — we're here to help. We've got a lot of different ways that we can provide assistance in getting you there, and we're doing it today with a number of customers. Lots of opportunity for you to adopt these capabilities within the platform, get some of those outcomes, and make sure that those AI pilots are a success.

 

Christine Morris

And Ian, we had a question from Sean. How do you envision CMP and Now Assist AI working together as an integrated model to drive value for cloud cost management within the platform?

 

Ian Cahall

Great question — really appreciate that. Number one, today there aren't any out-of-the-box skills or agents for cloud cost management. But going forward — and if you recall, SAM Enterprise is becoming SAM Prime. Historically, SAM Enterprise has been Software Asset Management and cloud cost management combined into one package. One of the things I anticipate happening in the near future, number one, is one or a few different cloud cost management AI skills for Now Assist — I think those will certainly be published. But in addition to that, I also anticipate that the SAM autonomous worker analyst will have some cloud capabilities. I don't know yet, because that won't release until the second half of the year. That's certainly a forward-looking statement, but I expect that those autonomous workers will also have cloud cost management capabilities inherently built within them.

In addition to that, obviously there's a whole world of opportunity when it comes to building custom AI agents to do specific things. So if you want an AI agent to handle your renewals or spin downs of cloud resources, for example, that capability will certainly exist. You just may have to take it from the 50-yard line to the finish line, if you will.

 

Christine Morris

All right. We're going to wrap up with one final poll. Have you evaluated your organization's AI readiness for IT Asset Management? A: yes, we're ready. B: yes, but we have some gaps. C: not yet, but planning to. D: we have no idea where to start.

Garrett says yes, but we have gaps. Jason is planning to get started, but not yet. Got some C's. I think we've all come to realize the power of AI — especially in the SAM area. If it can act as an expert in these areas — where you're just managing licenses and don't necessarily understand the breadth behind the technology that drives those things — the cost savings are just incredible. It's one of those things where it should fundamentally pay for itself.

With that, we are going to wrap up. If you're interested in hearing about our AI visible value assessment — it's not expensive, but it really helps on where to get started. The question around governance and risk is a big deal. You've got to start from scratch on what your governance is going to look like. You've got to get aligned with your security groups. Those are the things we can help you with. We've also got AI Spark offerings. If you're ready — let's do it for SAM — don't hesitate to reach out to us. We're always happy to help. And a final plug — if you are interested in some of our other MasterClass sessions, please go out and take a look.

We hope to — Ian and I will both be at Knowledge. I don't know, Kristen, if you have our booth number handy, but we'd love for you guys to stop by and say hi. We've got quite a few folks that have been with us from the very first CMDB MasterClass. That would be great to meet up. Booth 5520 — come stop by and say hello.

 

Ian Cahall

And I do have a minute or so to answer any standing questions if you have any.

 

Christine Morris

Sounds like it was quite helpful. As always — when you get the recording, you'll have links to us. You can always reach out if you want to have a further in-depth conversation with Ian on where do you go from here.

 

Ian Cahall

Thanks, everybody.

 

Christine Morris

Thanks, everyone.