One common use case for internal chatbots is tech support. This chatbot for help desk assists employees with technical problems they are having with internal systems.
The chatbot would normally be available as a web chat or embedded in one of the organization’s chat applications, such as Slack, Teams or Skype. Employees who are having system problems can then engage with the chatbot to get help.
- The first route to assistance is providing help with a problem that it was explicitly set up to deal with. In this case the chatbot would normally have a set of questions that it would ask to properly diagnose the problem. Once the problem has been identified the chatbot can recommend the solution and find out if the user’s problem is solved. This is the perfect use case for the chatbot, as it can provide immediate assistance and does not require any human involvement.
- The second route to assistance is using the knowledge base to provide assistance. The chatbot will guess which passages in articles in the knowledge base are relevant based on user questions and will provide these passages to the user in the hope that one of the passages is useful.
This functionality is similar to the way Google improves searches over time as users will reinforce the links between certain questions and certain passages over time, based on the user’s ultimate selection and possibly feedback. If possible, the chatbot could ask certain general questions before providing the link to be able to diagnose the problem better in the future. For example, it could ask the user which system this problem was associated with and whether it is a bug or a case of the user being stuck.
- The third route to assistance is human escalation. Again some intelligence may be performed by the chatbot in this case by associating the question to the relevant help desk (assuming there is more than one). It is vital that human support is available as a backup to the chatbot as it is unlikely that the majority of questions can be answered satisfactorily by the chatbot. The human agent can also be given the power to help the chatbot improve. For example, if a question comes through to the human agent, they can manually link that question to a given intent or answer so that the chatbot is able to deal with the question automatically in the future.
Ideally, the human agent is available in real-time on the same chat as the chatbot and can instantly intervene to sort out the problem. It may also be the case however that the chatbot needs to raise a ticket in the support system which support agents will address at a later point.
What this means is that the user would describe their problem in natural language and the chatbot would try to determine what the problem was from the user description using a natural language processing (NLP) algorithm. The chatbot would extract relevant parameters, such as the system in question, from the user’s statement using the NLP algorithm.
Once it has identified relevant parameters and the problem, the chatbot will ask the user further scripted questions to fully identify the problem. If it fails to identify the problem, the chatbot will search the knowledge base or escalate to a human as mentioned above.
- The chatbot can use screenshots to identify problems. A screenshot can not only help the support agent if the problem is escalated to a human, it can also be used by the AI to help identify the problem. Obviously in order for the AI to work effectively it needs plenty of examples of problems.
- The chatbot can be embedded in the software application itself and can detect when an error has happened. The user would not need to switch applications to report the problem and the chatbot could automatically provide additional information about the problem at hand, including a screenshot.
The ideal scenario is that employees have access to many different systems and services through the chatbot. This way they no longer have to figure out where to go to get a problem or query addressed. Even better is the scenario where the employees do tasks from within the chatbot. This can apply to doing things to resolve issues they are having. For example, an advanced tech support chatbot would allow employees to reset their password from within the chatbot.
Of course, ultimately, the success of the technical support chatbot will be determined by the quality of the implementation of the chatbot as this determines the value it adds to the employee versus the status quo. If the employees have confidence that it will add value, they will use it.
Disclaimer: We encourage our blog authors to give their personal opinions. The opinions expressed in this blog are therefore those of the authors. They do not necessarily reflect the opinions or views of Botpress as a company.