Video Transcription

Okay so, we have spent some time now looking at automation driven by ChatOps, looking at Slack and Microsoft Teams. But now it's time to take this concept to the next level. And we'll do that by integrating with fully fledged virtual agent platforms. So, I'm going to demonstrate this by running through our most recent deployment of this type of combined solution, as well as the associated use case.

Now, please note that I can't share the details of the virtual agent solution used here, so we've anonymized this content. But, of course, I can still step through the platform, the solution itself, and the use cases, and try to give you a flavor of exactly what was delivered.

So, first, let's take a look at a high-level diagram representing this combined solution. So, it's broken up into a number of different sections from left to right here. And on the left-hand side, you can see the entry point into the combined solution. So, an end user will access the virtual agent using the app either from their desktop, or from their mobile device. In fact, the virtual agent component in this solution can also be front-ended by existing solutions. So, we can integrate the system with your existing email, your existing ChatOps, your existing collaboration platform, or even custom portals. So, multiple entry points kind of moving towards omni-channel self-service, in this case. And, essentially, the user interacts with the virtual agent and states their request or whatever their reason is for interacting with the virtual agent.

The next step, then is to move to the virtual agent itself. So, what do we have here? Well, we have, first of all, the front end, we've spoken about that already. Second of all, we have a number of components that enhance the end user experience here. So, first of all, there are more than 1 billion entities, intents, phrases understood by the solution from day one. So, in terms of getting up and running, and ramping up with the virtual agent and automation combined solution very quickly this is, obviously, extremely powerful.

NLP, or natural language processing, is built in, so the virtual agent is able to natively parse the text, the questions, the queries, the commands entered by the end user and discover the intent. And it can do this in multiple different languages as well. And, from that point, there are a couple of different paths we can take. So, number one, if you see these arrows going downwards here, we can drive the end user into knowledge. So, easy access to knowledge and information available at their fingertips in your knowledge platform of choice. So, there are some examples given there in the diagram. But, more importantly, we can drive over here to the right-hand side, and we can use the intent discovered by the virtual agent to start triggering powerful backend automations and resolve actions.

Now, of course, there's a handover process here, there are outros going from left to right, and there are arrows going from right to left. So, this is a bi-directional integration in terms of the overall combined solution. We can trigger automations, the automations can communicate back to the virtual agent. So, in terms of the end-to-end interaction, you have this full back and forth from a conversational perspective. And then, on the right hand side, of course it's always worth mentioning that we could integrate seamlessly with your ITSM platform of choice while we're doing all of this. In this case, it happens to be HP Service Manager.

So why is this... or what is important to note here compared to what we've already seen when we looked at the ChatOps integrations with Teams and Slack? Well here, as I said, we've moved to the next level. So, we have a consumer like app experience. As I said, we have more than a billion phrases understood from day one, and this list is being continuously augmented. We can manage the full end-to-end interaction life cycle with the end user. And this, of course, leads to significant increase in employee adoption, reduced call volumes as you're, obviously, driving more towards self-service now. The ability to provide personalized answers to questions and issues, and guide users and employees through these complex processes. As well as, of course, easy access to knowledge and information. All of these things combined are extremely important and valuable.

Now, to get started with this use case, or this set of use cases, the entry point for each of them is exactly the same. So, we go and interact with the virtual agent, as I said, via desktop or mobile device. And we're presented with the virtual agent interface. And we have two ways to interact here. We can customize a list of common issues that appear as links on the screen to be clicked. Or we can type our question or our problem in the text box at the bottom. And, from this point onwards, the interaction has begun.

The first use case we delivered here was a very basic one, a password reset. And you can see it here represented on the screen, the end user requests that their password be reset. The virtual agent immediately understands the intent of this question, and verifies it with the end user. And then, the virtual agent initiates, in this case, an automation in Resolve Actions, which goes and actually performs the password reset itself, as well as some other validation, verification activities. So, I found the user account, the user account is enabled it's not locked. And then, ultimately, the password has been reset and can be shared via SMS. Can I help you with anything else? No, we're all good. We can move on to the next request or the next problem.

So a very, very, very basic use case to begin with, but you can already start to see the difference between the previous examples we showed and this one, the whole conversational aspect, the end-to-end interaction component, being able to jump in and out of this conversation and this interaction to drive backend automations to integrate with your key systems and your key applications.

Again, a similar example here but, in this case, the question being asked of the virtual agent is “I want to upgrade my phone,” or even potentially, “I want to upgrade my laptop, when will I be eligible to do that?” Again, the virtual agent understands that intent and asks the user, "Would you like me to go and check your eligibility?" Once they respond with, "Yes," again, we trigger that backend automation in the background, and now retrieving the details for your devices from some system within the backend application stack. We find the device, then we go and figure out when that eligibility date lies.

You can see actually the automation on the right-hand side that was associated with this end-to-end interaction here. You can see we're doing everything from creating a ticket to track this interaction, to retrieving the devices, and then looping through those devices to get the details to calculate the upgrade time, and then to provide that information back to the end user in the virtual agent, as well as updating the associated ticket, and potentially closing that ticket as well. So, the whole life cycle is covered here in terms of the end user interaction, as well as the backend process that needs to be executed.

Starting to get a little bit more involved and a little bit more complex now, in this case, the end user here wants to install an application on their laptop. Now, for this particular customer, there were two options here. Number one, they had a list or a catalog of approved applications, which were stored in SCCM. So, we trigger a backend automation to go and retrieve the applications in that catalog, and check if the requested application is part of that list. If it is then, again, the automation, we know we can jump back and get a confirmation from the end user. We can jump back into automation, go and gather the device, or the laptop associated with the user and initiate the software installation as it is pre-approved. The other option or flow here is that it's an unapproved application. So, if it doesn't fall within that pre-approved list, then we can simply go and request a ticket to gain approval for that user to install that application.

In terms of integrating with knowledge, as you saw in the very first slide at the beginning of this part of the demonstration, you have the option of flowing directly into knowledge sharing. So, going into your knowledge platform of choice and presenting the end user with the knowledge they need in order to resolve their issue. A lot of this is already built in to the virtual agent itself. And, of course, we can hook it up to your platform of choice as well to share this knowledge.

A more complex example here, where we have the user reporting their laptop is underperforming or running slowly. And, again, you can see this full end-to-end interaction across multiple screenshots here:

"My laptop is running slowly."
“Okay, I understand it's not performing. Would you like me to perform a diagnostic?"
"Yes."

Immediately the backend automation is executed and we're executing this broad health check on the end user computer. Once that happens, we go back to the virtual agent to get confirmation that we're presenting all of the outputs of that diagnostic. We go back to the end user to get confirmation that we can indeed go ahead and attempt the triage, the repair activities. And once they confirm, again, the automation can be driven to go and execute those steps.

Of course, all of this is backed up by the availability to introduce... or to include sufficient error handling across all of these processes, so that if there is an issue, for example, with a connection, or an integration, or access, or not enough details being provided, for example, then we can take an alternate path seamlessly without any disturbance. And we can let the end user know that something went wrong, we're creating a ticket. This will be followed up, “is there anything else I can do for you in the meantime?”