What is our primary use case?
We did a project for a client last year. We used the Paris release and did more than 15 use cases for the client with the Paris release. We also customized it a lot because the client wanted a different workflow for it. After that, San Diego and Tokyo releases came, but I've not done any virtual agent development with those versions. There've been no major changes in these newer versions. They just added extra functionalities for virtual agents. For example, they added extra languages for a multilingual bot.
It was mainly used for service desk help. The main use cases were related to incidents, service catalogs, and checking ticket status. We also did some integrations where the client wanted password resets through an external system. They were maintaining the data through a different system. When end-users came into ServiceNow Virtual Agent and said password issues, they were redirected to a third party where they were required to enter their unique user ID. With that, another system recognized it and did the needful, such as giving them the link, redirecting them, or giving them a KB article showing them the procedure for what they were asking.
How has it helped my organization?
It has been useful for mundane and repetitive user requests handled by the service desk. In the case of any laptop issues, password reset issues, or incidents that users were facing related to IT, instead of going to the service desk and talking to a service agent, they could place a request via a virtual agent. It was a guided, step-by-step procedure to help them log an incident.
It has also been helpful for catalog requests where an end-user could place a request for a device, such as a mobile phone or something else. Within the virtual agent, we didn't have complex steps. Our client wanted just a hyperlink so that the end-users were redirected to the service catalog. They would go to the portal and order it, but the initial conversation would start when an end-user came to the chatbot and typed in the request. That request was automatically understood by the machine, and it was redirected to a particular catalog or whatever the user is asking for. The end-users could also check the status of the raised IT tickets.
ServiceNow comes with a dashboard where performance analytics reports are generated. You get visuals of how a virtual agent is performing, which is helpful for the head of the service desk to analyze how well a feature is helping them. Dashboard reports help them understand the impact.
What is most valuable?
The ITSM topics were valuable. They reduced service desk usage and freed agents. They saved 40% to 50% of the time, and the agents were able to do other tasks instead of just replying to customers, replying to their emails, or logging the incidents themselves. That percentage of work was now done directly by end-users who were also happy because they were able to get the query response within seconds instead of waiting in a queue or waiting for an agent to reply. There was no delay in the first response. There was 50% or 60% less overhead on the agents, and we also saw customer satisfaction because we provided them with a survey. After a conversation ended, we would give end-users or customers a survey to rate their conversation with the virtual agent.
Virtual Agent is also a scoped application. ServiceNow has a really nice migration system.
What needs improvement?
While interacting with the virtual agent, sometimes, in the middle of the conversation, the end-users wanted to go to a live agent because the natural language model (NLU) of ServiceNow couldn't understand what the end-user was saying. There were some occurrences where users would just opt for the live agent support, and the virtual agent would directly route them to a live agent. However, there were some ServiceNow bugs that we had reported. There was also a known error where if a virtual agent wanted to connect to a live agent, the live agent was not able to monitor whether the end-user was still there in the chat or not. If the end-user left the chat, the agent was not able to know that. The agents were just waiting for two or three hours, and the conversation was not ended by the system as well. So, the problem was that the conversation was not ending. An end-user left the chat, but the agent didn't know if the user was there on the other end or not. This problem was fixed in the San Diego release.
The language or multilingual features where the machine was doing the whole thing was a little bit tricky. They have been improving the NLU models, but if we wanted to opt for a multilingual virtual agent, we had to train it for NLU properly. That was still a tricky part of ServiceNow. There were occurrences where the machine was not recognizing the topic correctly and not routing the users to the correct topic. It was specifically related to the Spanish language.
What do I think about the stability of the solution?
It's stable, and they have been improving it.
What do I think about the scalability of the solution?
It's scalable because they're also giving us the option to use models that are outside of ServiceNow. They give you a ServiceNow model, but they are also allowing you to use the IBM bot model, which is IBM Watson. So, you can integrate it with other third-party tools as well. If you want to use WhatsApp or Microsoft Teams as the channel for the end-users to communicate, and you don't want your users to go to the ServiceNow service portal and go to the virtual agent, you can do that as well. You can integrate it with Microsoft Teams, Slack, or WhatsApp. WhatsApp would be a custom integration, whereas, for Microsoft Teams and Slack, out-of-the-box integrations are available. Nowadays, ChatGPT is there. There is a possibility to integrate ChatGPT with a virtual agent. It would give really nice results. For the general queries of the users, it can be integrated with ChatGPT.
How are customer service and support?
Their response was quick, and they assigned us an Indian colleague who helped us to understand the actual issue. We recognized the issue, and they told us it is a known error. The client was not happy with that, but they told us that it'll be rectified soon, and they are going to release the patch to rectify the problem. That made it easy because it was a very quick response. I would rate them an eight out of ten.
How would you rate customer service and support?
Which solution did I use previously and why did I switch?
I've only worked with ServiceNow Virtual Agent. However, I have an understanding of other vendors and tools. They use chatbots. Chatbots and virtual agents are quite similar, but chatbots have a predefined workflow. There is no dynamic input, or they can't understand a dynamic input coming from an end-user. The flow is very static and specific to a particular process. Chatbots don't understand natural language like ServiceNow. They don't try to understand your query, and repeatedly, they give you the same replies, whereas ServiceNow Virtual Agent tries to understand, and if it's not able to recognize or identify your issue, it'll just ask you to talk to a live agent and route you to a live agent. It tries to understand the language. That's the plus point here.
How was the initial setup?
We had three instances. We used to develop in the Development (Dev) instance. For the user acceptance test (UAT), we would migrate it to the Test instance where the stakeholders did the testing. After they approved it, we would move it to the Pre-release or Pre-prod instance, which was a duplicate of the production system where nobody would test it. It was just to maintain a replica of the Production (Prod) instance. From Pre-prod, it would go to Prod. So, the release cycle they maintained was from Dev to Test, from Test to Pre-prod, and then from Pre-prod to Prod.
We used to do deployments via update sets. ServiceNow deals with update sets. So, the entire work we were doing in the Dev instance was captured via update sets, and we could migrate it via XML, which means we could just export it and import it to the other system. We could also just link the Dev instance to the Test instance, the Test instance to the Pre-prod instance, and the Pre-prod instance to the Prod instance. It was a chain of connections. From the Test instance, it would go to the Pre-prod system because we needed the UAT to be done, and whatever successfully reached the Pre-prod system would go to the Prod system. The Pre-prod and Prod systems had an out-of-the-box connection.
For the deployment, we didn't need ServiceNow's involvement. It was done by the team handling or doing the development. Usually, the project teams decide their deployment window to be on the weekends or Saturday nights. If it was a US project, then it would be nighttime in India. We had to give them the entire Excel sheet or plan for deployment, but all core activities were handled by the person handling the core ServiceNow operations. The people involved were a part of the release team. Two or three persons were enough for the deployment, and it was usually done within three to four hours. If there is a huge amount of data to be transferred, import would take 20 to 30 minutes. If there were a small number of topics or very small workflows, the import would take two to three minutes. Once you import, you just have to preview it and publish it, and it will come into the Production system. After that, you have to do just one test, which is a release test. So, all these activities were performed during the three to four hours deployment window.
In terms of maintenance, if you want to include some extra topics or modify existing workflows, you need an operations team. Every ServiceNow project that is being delivered to clients would have an operations team for enhancements. generally, two people would be enough for that. Other than that, there is no maintenance.
What's my experience with pricing, setup cost, and licensing?
I am a developer, so I've not been involved with its licensing, but I've read about it. Based on what I know, when you purchase ServiceNow ITSM, ServiceNow Virtual Agent comes out of the box with it. You don't have to pay extra for the ServiceNow Virtual Agent license. So, there's no extra cost. It also depends on the features and functionalities you are going for. There's ITSM and ITSM Pro, and it comes with ITSM Pro.
What other advice do I have?
I would advise starting with very simple workflows. You should first set up the live agent support and then go with very simple workflows, such as incident or IT status check. You should progress slowly. Go with the pilot release first with two or three topics, and when the users are getting used to it and are enjoying using the virtual agent, enhance the workflow. That will be helpful. The planning should be proper. Do a pilot project and then go with more and more topics once you're familiar with all the features. You can use the dashboard reports to see how much it has been utilized and how many times the users are using it.
They have been coming up with a lot of new features. They recently launched small talk topics. For instance, a user starts the conversation by saying that he is facing some issue, and the virtual agent recognizes it and starts a topic for raising an incident. To raise an incident, two things are important: the description and the urgency of the issue. Once they fill in those details, the system will log the issue and give them the ticket number created in the system. However, while filling that description and urgency, if the user asks the virtual agent how's the weather or something else, to address such things, they have come up with small talk topics so that NLU is able to recognize it. Of course, you need to build the model in a way that it understands that and gives a proper, valid reply, and then continues the flow, but it's an interesting feature. They have been improving it and making it a little bit smart. Let's see how it goes in the future.
As the developer of all the topics and as the end-user role that I played for the unit testing, I would rate ServiceNow Virtual Agent a seven out of ten.
Disclosure: My company has a business relationship with this vendor other than being a customer: Partner