How Smart Does a Knowledge Bot Need to Be?

By Doron Gower   |   February 21, 2018
How Smart Does a Knowledge Bot Need to Be
Now that we are in the summer this year, we can step back and take an assessment of 2018 customer experience trends and how they are impacting the marketplace. The industry has faced new challenges and developed a wealth of new approaches to overcome those challenges.

Bots are here to stay, and they will solve increasingly complex problems. According to Forbes, which based its article on a Quora question, bots are evolving. How smart your bot needs to be depends upon its purpose. Some developers think that a bot should have only one purpose, or job to fulfill, in order to build its knowledge base and become an expert in that one job.

For startups or SMEs, that may be a great starting point. For larger enterprises, that will not be enough. A knowledge bot must be able to reach a higher level of intelligence that goes beyond one purpose or job.

Developing Bots Is Not Enough

For information focused knowledge bots, development has been made easier with services such as API.AI, Wit.AI, and Watson Conversation. Knowledge bots are much easier to develop and deploy today, but its not enough. The time-consuming and difficult element to developing a knowledge bot is training the knowledge bot with the information it needs. As with a new business, initially, you should do one thing especially well. Your bot should be able to answer and act as the subject matter expert on the specific service it provides. It’s possible for a knowledge bot to attempt to tackle too much at the beginning. Trying to master all knowledge about a specific service or domain is a large task. By focusing on purpose, eventually, your bot will be smart enough.

The Minimum Viable Knowledge

In product development, the beginning is minimum viable product (MVP). With bot development, the beginning is minimum viable knowledge (MVK). By starting with one specific element of knowledge and then building upon it, the bot’s purpose will merge with its knowledge. By defining the purpose, you then develop the scope of its knowledge. This then determines how much training the bot will need.

As individuals interact with the bot, you will gather more insight into the particular type of user interfacing with it and the questions they ask. You can then utilize those insights to give further training to your bot. This works especially well for content and information focused bots.

As the market grows for bots, you may actually be able to conduct something akin to an interview with the prospective bot prior to choosing the one best suited for your needs. Having a bot with a well-defined purpose and that is the subject matter expert on its domain will help determine just how smart your bot needs to be. This improves the chances of success for both the bot and your company. Service levels should improve as the bot gains in knowledge, and that all begins with answering the question: “What is its purpose.”

Schedule a FREE consultation with a BOT master of KMS ligthhouse to learn how your bot can be trained for successful customer expereince: