5 Questions You Should Ask ServiceNow About Their Virtual Agent at Knowledge 20 ( Plus 1 Bonus Question)
If you’re a ServiceNow customer, chances are that you missed going to the annual Knowledge conference in Las Vegas last week because it was always a great place to network. As you know, due to COVID-19, they moved their event online throughout May. Stretching over an entire month makes it harder to navigate, but there are always important sessions that I bet you will still attend.
Since the work from home mandates struck, help desk automation has gone from a nice to have to become an imperative for IT service management (ITSM) to ensure workforce productivity. With that in mind, you may be opting into sessions that discuss their virtual agent. For those of you that do, I thought I would provide you with a list of things to consider and ask ServiceNow.
- Will I need separate chatbots for each department?
Answer: Yes, you will need a separate chatbot for each department.
ServiceNow approach: Maybe you are thinking you can start off with a virtual assistant for IT and then expand into other areas. It’s a good idea because after getting the hang of using one virtual assistant, employees will be reluctant to learn another one. Not only that, having just one place to go means that employees are spared the burden of memorizing the location of multiple chatbots, or having to guess which one is the right one. Is a lost ID badge the IT chatbot, the Security one, or maybe Facilities?
The ServiceNow Virtual Agent, along with the natural language understanding (NLU) components, are designed to be deployed separately for each department. While this allows each department to pursue separate projects and timelines, this approach forces users to figure out what is the right chatbot to use for each problem that they have.
Espressive approach: Espressive Barista already recognizes 80-85% of what employees might ask across any department. Barista is purposely designed to be a single place for all employees to go to get the answer to any question, which is exactly what employees would expect.
- Is your virtual agent one product that I am buying and deploying, or many?
Answer: Many – ServiceNow Virtual Agent plus Studio plus Agent Assist plus Orchestration.
ServiceNow approach: Purchasing an enterprise grade virtual agent from ServiceNow is not inexpensive, and neither is the huge investment in internal project resources needed to make it work—so it’s best to know exactly what you are getting when selecting this product. Like other products in the ServiceNow portfolio, their virtual agent is a stand-alone product that can function independently and in conjunction with other ServiceNow applications. With their virtual agent you can build conversation flows and other actions users can perform. However, if you want the virtual agent to recognize what employees are asking using natural language, that is a different product.
For a virtual agent that understands what employees are saying, you will need ServiceNow Studio to build your NLU models. If you want to use machine learning to populate the ticket values using AI in order to save time and lower your MTTR—well, that’s a different product called Agent Assist. Do you want the virtual agent to look up data and perform tasks in other systems? Then your team better be familiar with ServiceNow Orchestration. Unfortunately, all of the skillsets, scripting language, and user interfaces are different for all of these products.
Espressive approach: Barista was designed from the ground up as a platform for conversational AI. That means you have everything you need in one place—and one price! This includes conversation design, language modeling, testing, integration, machine learning, and knowledge article ingestion.
- Will the virtual assistant use AI to get smarter over time?
Answer: No, the ServiceNow Virtual Agent does not have a mechanism to get smarter on its own.
ServiceNow approach: One of the apparent advantages of AI is the ability to learn over time. The expectation is that the AI will observe feedback or corrections from the data and retrain itself to provide a more accurate result. The ServiceNow Virtual Agent does not have any mechanism to get smarter on its own. All improvements must be done by your data science team—this includes improving the language recognition (i.e., matching), or expanding the language model so that it understands new content. Getting the virtual agent to ask a different question or a new question is also a manual exercise.
Espressive approach: In contrast, Barista does get smarter every day because Barista synchronizes with the Employee Language Cloud (ELC), which continually fine-tunes and expands the language model. Espressive uses feedback from all customers in order to make changes centrally, which instantly improves outcomes for everyone. One of the biggest disadvantages of AI projects is not getting enough data to drive improvements. Espressive has solved this problem by leveraging massive amounts of data from the entire customer ecosystem.
- Does the virtual agent come with most of the content that I need to get started?
Answer: No. It has a couple dozen example topics that can be customized – that is not content.
ServiceNow approach: Content refers to both what topics the virtual agent understands and the responses or answers it has for employee questions. Simply put, if a virtual agent is loaded with relevant content, it will be perceived by employees as an intelligent assistant that is worth coming back to. However, the opposite is true if the virtual agent only understands a few topics. Employees will not find what they are looking for and possibly never come back. This is why the concept of starting small and expanding doesn’t really work with employee facing chatbots.
The ServiceNow Virtual Agent does not have real starting content. Instead, it provides a couple dozen example topics that you can customize to your needs. While this is helpful and better than nothing, it does not come close to the thousands of topics employees actually ask about for both IT and non-IT related questions. This means your team will always be chasing behind what employees are asking for, which results in poor adoption.
Espressive approach: Barista has been helping employees for years and the Employee Language Cloud (ELC) understands over 750M phrases across the majority of topics that you will need. This means that you will not have to build your own topics or language models. You will not have to start from scratch and figure out on your own what employees care about—all of this has been done for you! The ELC even includes hundreds of researched answers for popular tools. These answers are kept up to date by a professional content team, so your team can focus on those answers that are specific to your organization.
- Will the virtual agent automatically leverage my Service Catalog and Knowledge Articles?
Answer: No. The virtual agent does not convert catalog items into a conversational format.
ServiceNow approach: You would think that, because the virtual agent is on the ServiceNow platform, there would be huge advantage due to “built in integration” to ServiceNow products, like the Service Catalog and Knowledge. Sadly, this is not the case. The virtual agent does not convert catalog items into a conversational format to take advantage of the chat experience. Instead, you will be doing a lot of work just figuring out how to link each catalog item and launch the user into the portal page with a different experience. The virtual agent also doesn’t provide any AI tooling (i.e., natural language understanding or machine learning) to ingest your knowledge data and provide employees with the most relevant parts of a knowledge article.
Espressive approach: If you want a high level of integration that just plugs into ServiceNow with very little effort, look at Espressive Barista. Items in your Service Catalog are automatically converted into virtual conversations with your catalog variables becoming the questions that Barista asks the employees. After chatting with the employee, Barista will create both the Service Request and the Request Item exactly like you would expect, so that the appropriate workflows and approvals are started. With Knowledge, your employee facing articles are automatically read and converted into a more accurate language model using machine learning. When employees ask Barista questions, Barista is checking for both easy FAQ responses as well as the most relevant sections across all of those KBs. Any content changes made in knowledge are instantly updated in Espressive after those changes are approved. Using different languages? No problem, Barista will automatically show the user the article that matches the language the employee is speaking.
- Can I leverage the same reporting, updating, and testing framework with your virtual agent as the rest of my ServiceNow applications?
Answer: No. The analysis tools are missing as well as adequate tools for testing in sub prod.
ServiceNow approach: Let’s first look at reporting, which is one of the key strengths of the ServiceNow platform. One would assume you would have an easy way to report on the questions that employees are asking to analyze what is popular or not popular, what are the overall trends in employee demands, etc. Reporting with their virtual agent is actually quite different and not what you would expect. While some of the raw data is there, the analysis tools are missing, requiring you to move the data to another tool or build your own tool to make the data meaningful. What about tools for testing changes in sub-prod and moving approved changes to the production environment? Unfortunately, there is even less tooling for ServiceNow customers using their virtual agent or the NLU Studio to control these changes.
After changing a language component, like understanding a new employee question, there is no visibility on how that change might adversely affect the accuracy of all of your other topics and questions. You could create your own manual test and amass a library of test phrases, but this would be a nightmare to process after each change.
Espressive approach: As a mature AI platform, Barista has everything you would expect out of an enterprise grade solution. This includes reporting and analytics that let you drive insight from employee activity. Barista has a built-in testing framework that allows you to run automated tests that identify any regressions that occur from changes. Barista also has specific controls that let you migrate all changes or just specific changes from sub-prod to production environments.
At the end of the day, any good marketer or salesperson can spin the truth to make their product appear to be the best. While I’m a pretty reliable resource as the first ever product manager at ServiceNow and now the chief product officer for Espressive, it’s always better to see for yourself.
After learning about their virtual agent, please take time to learn about Espressive Barista, our virtual support agent and request a demo here. Be sure to request to have me on the call! I’d love to join.
In addition, you might want to check out our blog on 5 Reasons Why ServiceNow Customers Select Espressive Barista.