top of page

ChatGPT in Knowledge Management and Service Desks

1 May 2024
Sagit Salmon
A hand holding a glowing screen

ChatGPT burst into our lives in November 2022. Since then, it has been challenging organizations and causing them to think about how they can leverage it, how it can integrate with existing technological tools and work methods, and how it can help streamline work and achieve a competitive advantage.

We, as knowledge managers, are constantly thinking about how it can serve as a tool for us, particularly in knowledge management work in service desks, both in the knowledge management team and in service desks.

How can the knowledge management team utilize ChatGPT?

At the core of the team's activities - editing and providing content accessibility. ChatGPT can be used to get assistance at the start of an activity, and then validate the proposal with the relevant audience, or vice versa - formulate a proposal and use ChatGPT to refine it and get additional ideas that haven't come up yet.

Here are some possible uses:

  • Defining a topic tree (sample query: what are the 10 topics that should be covered in a knowledge management system in service desk X?)

  • Defining a template for a selected topic (sample query: suggest a template for writing information about a product in service desk X)

  • Organizing content in a template (sample query: help me organize the attached content according to the following chapter headings... In this case, you need to attach the text and the chapter headings of the template)

  • Content editing: writing tailored to the audience, identifying key topics, filtering out minor topics, summarizing, and organizing according to the work process.

The range of ideas is expanding as the tool develops, but this is certainly a good start.

As for the use of ChatGPT in service desks, technology companies, and organizations are working on integrating ChatGPT into existing operational and knowledge management systems, which will form the basis of the information on which the chat is based.

In the long run, people are already talking about a bot that will replace a representative and talk to the customer (there are organizations where this already exists).

Why is this mostly considered a long-term prospect? Customers need a high level of trust in the entity handling them. The chat still struggles to process information and use it logically and critically, and it cannot identify errors.

Therefore, at this stage, the proposed uses are more modest, with the main purpose being to use large amounts of information to provide better service to customers and improve the customer experience.

Here are a few ideas:

  1. Structured search engine queries:

    Formulate an initial set of queries that can assist representatives at the start of their work and train the tool. Later, employees will learn to work with the tool and know how to ask questions, and in parallel, the tool will learn the work habits and key questions.

    It's advisable to set some default conditions for the tool to display answers, so they don't need to be defined in each query separately, such as up to 200 words, up to 4 paragraphs, up to 3 topics, etc.

  2. Written communication: Use the tool to provide written responses (e.g., email service or on WhatsApp and social networks), based on previous responses and the existing knowledge base.

    1. Examples of possible activities:

      1. Using generic responses to recurring questions

      2. Adapting the response style to the customer (message language, message length, and even the customer's mood)

      3. Identifying previous responses on the topic and formulating a personalized response

      4. Responding to customer reviews posted on social media

      5. Assisting in sales processes (answering questions from potential customers, customizing sales offers)

  3. Knowledge documentation: Assisting in writing and documenting knowledge, a challenging task that requires time and thought:

    1. Documenting inquiries and formulating articles derived from them

    2. Accurate and high-quality description of issues transferred between different entities in the company

    3. Writing user manuals and solutions for customer use

Even here, there are still many ideas, and even if not all of them can be implemented yet, it's only a matter of time.

What is important to ensure when integrating ChatGPT in an organization?

  • Data security - preventing internal information and customer data leakage

  • Adapting the tools to the organizational systems and controlling their operation

  • Training employees to use the tools efficiently

  • Combined work of the tool with human resources - to maintain credibility with customers and prevent incorrect answers (the tool is based solely on existing information and may provide an incorrect answer when there is no suitable information in the texts)

In summary, good customer service is not only a good experience; it's also a response and solution to problems.

ChatGPT serves as a significant tool, mainly in responding to simple and generic issues.

Using it can, on one hand, shorten the time spent handling customers, and on the other hand, allow the existing human workforce to focus on quality solutions and providing value to customers.

And when will the chat replace customer service representatives to the satisfaction of customers? We're waiting curiously and patiently.


bottom of page