5 Tips to Deliver Agentic IT Service in Slack | Dreamforce 2025

My name is Julian Walker. I'm on the service cloud product marketing team at Salesforce. Me and my team have had the pleasure of bringing AgentForce IT service to market the past couple of months and learning from so many of you and different discussions. And I'm joined by my colleague Caleb. He has the pleasure of going to meet so many of you customers face to face and show you the value that AgentForce IT service can provide. So without further ado, let's just hop into it. So we are really at an inflection point when it comes to this era that we're in. We're stepping into this agentic era and what this means for IT service is we're really moving towards a more conversational, agentic type of experience. So tell me if this sounds familiar. If you have an issue or your employees have an issue, they go to the portal, they, they type in their question, they get a knowledge article, they fill out a form, they get a ticket and then that ticket goes to an approval to a manager. And somewhere along the way they get a response from IT that says we'll respond in the next business day. And that may or may not be the case. That's just the standard message that everyone gets. If it's something urgent, you might get a response a little bit sooner. But the reality is if I had something that was really urgent that I needed to get done, I'm sweating at that point. If I needed to have that done by end of day and I'm getting a response the next day, that's a problem. And you know what, that's why we're moving into this new era of agentic IT service where things are AI agent first, they're conversational, they're happening right in Slack where your teams are working every single day. And we're able to be a little bit more proactive than we have ever before in order to reduce costs. So that's exactly why we have introduced AgentForce IT service. It's so that your teams can resolve issues 24, 7 conversationally. So that means in the flow of work and that's going to be a big benefit for your business. Ultimately, it's faster resolutions because your employees are getting answers right within Slack with AI agents. Secondly, it's more productivity for your IT teams. We're automating end to end processes so they can get help along the way. And number three, we're reducing IT cost. IT is not sitting in a silo. We have all of your IT tools on one platform. Your AI agents, your CMDBs, your IT service desk, your AI agents once again right in one place. And lastly, you're able to delight your employees at the end of the day. It's a really frustrating experience when you have to go from portal to email to all these different places where. But if you can do everything that you need to right in the heart of Slack, that's a seamless experience. So I promised you five tips. So we're gonna go really quickly through five tips on how you can deliver agentic IT service right inside of Slack. And then we're gonna get to what you've really been waiting for, the demo. So let me go through these tips really quickly with you. So number one, deliver faster resolutions with AI agents for employees. What this means we're going to give an AI agent to all of your employees in that scenario that I mentioned, like if I needed to get Photoshop, I'm a marketer. If I needed to get Photoshop to finish a project at the very end of the day in that old scenario, I would have had to go to that portal, go through that full process, get that ticket, get my software the next business day or maybe even later because it maybe wasn't super urgent. But in this new world, all I'm going to do is go in Slack, start up a conversation with my employee agent and boom, within minutes I have access to Adobe Photoshop. That's pretty cool. Obviously with the right permissions in place, we can still do approvals and all those things on the back end to make sure you're just not giving me software, any software that I ask for, but that's all part of the process. Number two, improve productivity with AI agents for your IT teams. So we've given every employee that employee side of the agent, but your IT teams eventually are going to start getting the really complex cases. And that's because something I learned From Gartner recently, 80% of issues at your company are going to be resolved without human intervention in the next five years. That's huge. So what that means for your IT teams is they're gonna be dealing with the really complex issues that come through their queues and, and they're going to need some assistance. They can't always be able to reach out to their manager on every issue. An AI agent can be there right by their side helping them every step of the way. So what does this kind of look like for your IT team? Well, we have over 25 pre built agents that we're going to be delivering to your IT teams over the course of the introduction of this product, imagine a root cause analysis agent instead of trying to figure out what the underlying issue is for something all on your own. What if Agent Force could be there right in Slack and I can just say, hey, what's going on with these issues? It can dig through all of the conversational data, all the ticket data and all the past incidents to figure out what's going wrong, make a recommendation, and maybe even give me some next steps to follow as well. Number three, automate with workflows for your AI agents and your IT service desktop. So we all know that approvals are a really big thing. I mentioned that in my last example. Now imagine if your approvals were just as easy as a click of a button. You don't have to go through an email chain or a back and forth with a manager to find more information with the employee, things like that. Well, we're delivering over 100 workflows with this solution as well, and approvals is just one of them. So the way that that manager, in my scenario of getting Photoshop was able to approve so quickly is because what happens, they get a notification in Slack, they open up that message, it has all the context that they need. The user that's requesting it, the software that's being requested, a business justification right there. And they have two options, approve or deny. Once they click approve, it immediately sends things off on the backend and that AI agent performs. Provision that software to that employee. Super simple. Now number three, extract enterprise context from Slack for your CMDB and your service graph. As I mentioned before, with AgentForce IT service, you have the CMDB, you have the service graph all in one place. Now we all know Slack, you have billions of conversations happening in there every single day. Some people are problem solving some people, some people are venting about their issues. Well, what if you could use all of that context to inform the CIs that are baked into your CMDB? Imagine you have an incident that's happening. It automatically kicks off an incident. There's a resolution that's happened, it's closed. Well, wouldn't it be nice if you could just pull all of that information directly into the cmdb? From Slack? Well, that's exactly what you can do. It can update right along the way so that your problem managers can have everything that they need every step of the way to not only resolve that current issue, but to resolve future issues as well. All right, number five, let's resolve complex issues by swarming with AI agents and subject matter experts. So I don't know if you guys have seen this use case before, but it's been around Salesforce for a while. It's called Slack Swarming where you bring experts into one channel in order to resolve something very quickly. Now with AgentForce and AgentForce IT service that can be done automatically as soon as we, you know, incidents can be a real drain on productivity. It causes a lot of chaos across the company if a site goes down, things like that. Now imagine if your site goes down and AgentForce can automatically create a channel with the right experts in it to resolve that problem. And guess what? AgentForce is also going to be in that channel that to give some recommendations along the way pulling from all of the other previous incidents that it's resolved in the past and also using any conversational data that's been happening there as well. And now imagine your CIO is wanting to hop in and see what's going on with this incident cause it's taking a little bit to resolve. If they hop into that channel, all they have to do is ask agentforce, get me caught up and it's going to give them a quick summary of all the things that have happened up until that point so. So that they can resolve it even faster. Pretty cool. Okay, now enough of me talking. I know what you really want to see is the demo. So Caleb, go ahead and take it away.

Thank you Julian. Good morning everybody. My name is Caleb Steen. I lead our solution engineering team for our brand new AgentForce for IT service solution and ready to jump into our demo. So I'm gonna jump into the role of Katie Holmes. Katie's an employee here at the organization. She's fairly new and she's just got a brand new machine. She needs to start to walk through her setup process. And so she's got access to Slack and so that's going to be her home base for the rest of our journey here today. And so when she opens up her Slack, she'll notice down here at the bottom her team is actually deployed. You'll see in the bottom left hand corner this IT help desk agent where she can now start to engage and ask it certain prompts, certain questions, even automate certain processes like we'll see here in just a moment. So she's going to say things like I need access to vpn. And it's at this point right to Julian's point a moment ago, we can start to now facilitate a guided experience for her to that's going to a validate this right information. So we're deploying this in the right places and then at the same time ensuring that this is gonna be done so in a way that it takes that burden off of her plate and creates efficiency for her IT team back behind the scenes. So in this case, the agent wants to validate. These are the current assets machines we understand that you have against your profile. Does this sound correct? Yes, that is. And as we go through this intake process, we will make it very seamless, very straightforward and very pointed. So Katie knows exactly what she needs to provide and what to expect. And in this case, this will be for my MacBook Pro. We'll go ahead and continue the prompt. And once the agent gets to a point where it has enough information to complete this particular action, it's going to go ahead and facilitate this case creation on our behalf. And as we can see, that service request has automatically been created and we can now see this pop up in the context of our IT team. So when our IT team receives that case, we're going to jump to our console here just for a moment. This is where we can now start to drive in. We'll go ahead and refresh, see that brand new case number from Katie just a moment ago and they can start to familiarize themselves with the case details, bring in any key party members that they may need to. But in this case, looks like we have a couple more steps that we need to collaborate and communicate with Katie before we get to that step. So Dave here on the IT fulfillment side will say, hi, Katie, looks like you need to actually install some drivers, A new driver. Mind if I push that for you? But right back in Slack because we now have this collaboration, this record channel against that service request that we saw just a moment ago. Katie's gonna have this brand new record channel pop up for her where she can now dig in, start to communicate with her IT fulfillment team right here in that same context. So no more are we gonna have to jump from platform to platform or UX to ux? We can now start to have this conversation right here inside of that continuous workflow. Right. So Katie's feeling good, she's got VPN set up, she's got the right drivers installed now. And now her VPN is gonna be pushed to her machine in a really effective way. So let's jump ahead in the story a little bit. It's been about a couple weeks. Katie's doing great. She's really enjoying this new organization. And all of a sudden she runs into an issue. She can't access that brand new vpn. So she wants to go right back to her agent. And for this case, we're gonna go ahead and start a brand new conversation. And Katie can now say things along the lines of, I am facing issues. Accessing vpn, Please create me a ticket. And it's at this point, depending on those processes, those workflows that you have defined, those instructions you provided to the agent, the agent's now going to come back and provide that very pointed and appropriate response. In this case, we've asked and instruct the agent to ensure that before we create that ticket, we want to take our employees through a guided troubleshooting experience. We can now provide this information directly from our policies, from our knowledge articles, all. All this wonderful information right here at the fingertips. And so as we're working our way through, Katie can now say, that didn't work. Please help. This is urgent as Dreamforce is this week. And so at this point, now that the agent recognizes that we haven't found resolution, this is an underlying issue that needs to need some assistance. It's going to go ahead and automate the creation of that incident, just like we saw from the case a moment ago. So we've created this incident and now we're going to switch gears and jump into the role of a brand new Persona, Dave, who is one of our IT fulfillers. So let's go ahead and open up Dave's Slack profile. Dave here logs in for the day. First place he likes to do here is go to his IT support channel. And the reason being, it gives him a very clean insight into all those open incidents that have started to come through over the course of his time offline. So he now has a quick, clean view of all the various incidents records that have been assigned to him. He can get a quick, high level bird's eye view of what's actually been occurring across those various incidents. But instead of having to go and hunt and peck through each single one of these records of these incidents, he can also engage with Agent Force to help get him to a level of more productivity and more efficiency. So he's going to say things like, you know, give me the rundown on all my open incidents. And it's at this point, Agent Force can now look across all those various assigned incidents in his bag, anything else that might be happening across the organization to give him a full summary and analysis of what's been happening across his organization in relation to the various incidents that he's focused on. So he gets the specific summarized view, even getting a little bit of a high level understanding of that root cause Analysis. But Dave realizes he needs a little bit more information before he's going to be able to start to propose a resolution. So he can actually go right here from the same record channel in Slack and open up the actual individual incident itself, where he can now review information. He can also manage and maintain an update, whether it's going through approval processes like Julian mentioned, whether it's updating key fields or attributes about this particular incident. But in this case, he needs a little bit more context. So he's going to go ahead and jump right into his console view that we saw just a moment ago directly to that specific incident. And this is now where, as Dave continues to work his way through, familiarize himself with all these various aspects based on the volume of these incidents he's seeing, he's realizing this is probably not a minor incident. We actually need to promote this to a major incident. So he's going to go ahead and fire off that workflow quickly, hit that button up here on the top, right, and that's going to go ahead and start that process. And we have this very specific automation and workflow tied to this, where we've now created a swarm behind the scenes, looping in all the appropriate experts and resources so we can have that subject matter expertise. As Dave is working his way through, however, this is where he can also start to gain some additional insights from tools like our proactive assistance, where he can now start to see were there any changes that were automatically made on his behalf. Right. Was this incident misappropriately categorized? Were there issues with the data itself, even looking at things like the root cause analysis potentially that agent forces is now generating for him, but also when we think about our data management being able to associate other similar incidents to that, so we have an appropriate grouping, we'll have a full view and and lens into what those potential issues might be. The other aspect, he also wants a better idea of how these various CIs are mapping into each other and what those impacted aspects and areas of his organization on the IT side might be. So related directly to this incident, we've also brought in this service map where he can now have a clean view and lens to understand the scope of what this potential issue might be impacting. So starting from the actual VPN service across all its various gateways, any other types of clients that might be impacted, even down to the server level, so again, giving him a 360 degree view of his entire network, all of his various assets that might be impacted, so he can come up and show Prepared as he jumps back into Slack. However, what we'll see here is he has this brand new channel that was created from that swarm I mentioned a moment ago. And this is where we can start to see that our collaboration agent has also stepped in, is going to give him insights to essentially where he needs to be focused to catch him up on maybe these particular issues might be. But you'll also notice automatically we've had some of those subject matter experts already added. So Julia as an example, as well as Zoe, who is one of our VPN specialists. So let's go ahead and loop Zoe into the conversation. So Dave right here can say, zoe, please take a look. This issue seems to be pretty pervasive. It's a big word on a Wednesday morning. All right, there we go. Seems to be pervasive. So we get that insights, Right. We can now start to have this conversation. Zoe's immediately ready to step in and start to collaborate with Dave, with Julia to help ensure that we can find resolution, not just to serve Katie, but the rest of our other impacted employees as well. So as we can see here in real time, we're starting to have this very tight conversation. Zoe's bringing in her expertise and she's starting to do some of her own research. And we now have that historical view, that conversation, and working towards that resolution. And it looks like she's always actually come back with a proposed solution. Right. As an immediate fix, let's deploy a new patch to revert back based on a recent upgrade that may have failed or provided some challenges. So let's go right back to Agent. Right. We're still in Dave's shoes. Dave is our IT fulfiller and we can quickly go and identify that same conversation we saw just a moment ago. And now from Dave's point of view, he can now start to provide some of those insights and drive towards that resolution. So in this case, we can now correspond with the agents say seems to be an issue with a recent security patch. Please revert back to the latest. Previous version and resolve our incidents. Right. Because it's one thing for us to actually go and push that security issue, but we also want to ensure that the agent has the ability and the agility to go and actually resolve all our underlying records behind the scenes so that we can have a clean shop and make sure our records are up to date. We can manage and we can report that up to our affected stakeholders as we see fit. So in this case, the agent understands our request provides some additional insights and opportunities here. And at that Point we say, yes, we do want to proceed, that's going to go ahead, push that patch to resolve that issue we've had from our security aspect that's actually impacting our vpn. And at that same time, as Dave jumps back to his IT support channel, to that incident list view, he can now start to see exactly the new change and all those various aspects of right here inside of his individual flow, giving him that visibility, that confidence and that trust that he now has an experience for his employees and he's able to be more efficient and move on to that next high priority aspect in which he's engaging. But also, let's step out of Dave's shoes just for a moment. I wanna step in the role of our administrator and this is where they can start to really understand and make how easy it is to deploy agents to Slack. And so in this case I'm gonna go to my Agent four section over here on the left hand side and we can now see when I go to my management screen, all the various agents that either a have been pushed from our agent builder that are ready for me to review and then deploy to Slack for my employees, for my IT fulfillers or whoever that end user might be, but also be able to manage and maintain all of those other agents that have been deployed against the same Slack experience, again giving me control, visibility and making it very seamless and easy to take all those workflows, all those instructions and deploy that to my employee base where they work. Right. In this case Slack specifically. And from an agent builder perspective, you might be familiar with this as you've gone through and built any agents in your own environments or any trailhead environments as well. Well, not only can you of course define those key topics, those data libraries such as knowledge articles or any key pieces of information that are important for our agent to be successful, but you can also define those very specific connection points, right? So whether wherever you want, you essentially meet those particular employees in their channel, whether it's in your portal, whether it's in teams, whether it's in Slack, we now have the ability to create that connection as you see here down at the bottom. And it makes it very seamless and very easy for you to create that, deploy that, and then like we saw just a moment ago, actually go through that review and approve that to successfully deploy that back to your employee base in Slack as well. So with that, well, that's where we'll conclude the demo. Thank you so much for the time today. As a call to action, please check out our keynote here. It's later this afternoon at 3:30. You'll see a lot more of what we saw today, but even a lot more exciting other additional capabilities when you think about the entirety of the solution outside of just the context of Slack. And also feel free to grab this QR code where you now be. Looks like we lost my slide there, but we also have the ability to go and see a little bit more and have a little more conversation. Some of our updates and some key news releases as well as we go GA in the near future. But thank you all for the time.