Chatbots have received a lot of attention over the past few years, but have generated mixed opinions. It’s hard to deny the value of an intelligent agent that can automate conversations with humans. But many agree that chatbots have been over-hyped.
Knowledge Assistants represent a new breed of bot that focuses on smart informational retrieval without requiring custom, pre-configured dialogs.
In this article, we will define chatbots and Knowledge Assistants, provide examples, and share 10 differences between them.
What is a chatbot?
Common internet definition for the chatbot:
“A chatbot (sometimes referred to as a chatterbot) is programming that simulates the conversation or “chatter” of a human being through text or voice interactions.” (Wikipedia)
The definition of a chatbot clearly explains why chatbots don’t meet user expectations. Expecting a chatbot to simulate a human conversation, simply put – is not possible yet.
Chatbots have been successful automating very specific simple tasks, but humans still have to jump in and take over for real conversations. The value of a chatbot is realized when time-consuming workflows that require a human, can be automated.
Example of chatbot: Drift
DriftBot by Drift is a popular choice to help you qualify your leads, increase website conversions and automatically schedule meetings with a chatbot. When needed, DriftBot will route your users to a human through real-time chat. DriftBot is a great example where the chatbot provides 24/7 assistance and successfully automates a process that previously required a sales person. Nevertheless, DriftBot has well defined boundaries, and Drift acknowledges the need for human conversations.
What is a Knowledge Assistant?
Related to chatbots and similar in form, but very different in behavior is a Knowledge Assistant. Chatbots excel at automating specific workflows, such as buying a product or scheduling a demo. Knowledge assistants serve an open-ended goal of helping users find information faster.
A Knowledge Assistant is a bot that is deeply integrated with a knowledge base, such as a website and is able to resolve user questions without having to be specifically trained to do so. A Knowledge Assistant requires an end-to-end pipeline to retrieve the best possible piece of content buried in a knowledge base.
Given information is messy and unstructured, knowledge assistants require more advanced technology in the form of Natural Language Processing.
Specifically, knowledge assistants need to excel in three areas:
- Understanding the user question.
- Retrieving the best possible knowledge assets for the question.
- Finding the specific answer within those knowledge assets. Which becomes a major data science problem.
Example of Knowledge Assistant: Frase
Frase is a Knowledge Assistant that ingests your entire website content and allows users to find information through a Question Answering experience. Frase leverages an “Answer Engine” to deliver specific answers (versus a list of links).
Chatbots and Knowledge Assistants are both very useful but in different contexts.
10 differences between chatbots and Knowledge Assistants:
Chatbots follow decision trees. Knowledge assistants are open-ended.
2. Goal setting:
Chatbots have very specific goals, as defined by the admin. Knowledge Assistants can have varying goals depending on the user intent.
3. Search capabilities:
Chatbots are not search engines. Knowledge Assistants are a kind of search engine specialized in retrieving answers – otherwise known as an “answer engine”.
4. Structured data:
Chatbots require structured data to resolve questions, such as FAQs. Knowledge Assistants excel at working with fully unstructured data, such as long-form documents.
5. Manual setup:
Chatbots require manual setup by an admin (for example: to create dialog content). Knowledge Assistants only require access to a knowledge base and there is no manual setup.
6. Dialog type:
Chatbots are transactional. Knowledge assistants are informational.
Chatbots are integrated with tools or platforms (such as marketing and sales automation tools). Knowledge assistants are integrated with knowledge (such as a website or intranet).
Chatbots don’t try to reason when providing options to the user. Knowledge assistants try to find the best possible response on their own.
9. User experience:
Chatbots usually start a conversation with qualifying questions. Knowledge assistants immediately give the user control.
Chatbots are usually text-based. Knowledge assistants adapt to more flexible environments, such as voice.
When considering whether or not to implement a bot experience on your website or other digital platforms, you need to consider what your end goal is.
- Chatbots are best suited for jobs where your AI agent will live in a highly structured content environment such as an FAQ or if the chatbots job is to complete very specific workflows such as scheduling a meeting.
- Knowledge Assistants are the best option if your AI agent will live in a highly unstructured content environment such as a repository of long documents, a website, or any kind of messy knowledge base.