What is a Chatbot?
Chatbot - benefits
increasing the efficiency of the customer service department,
providing 24-hour customer service,
a reduction in maintenance process costs - by up to half,
relieving employees of the burden of schematic work,
quickly satisfying the need for information - customers and employees,
standardisation of processes,
increased levels of customer and employee engagement,
building customer relationships,
improving the company's image.
Chatbot functions
The chatbot is intended to have one key functionality – to answer the user’s questions. The user types a question in a box and the online chatbot, acting on the basis of advanced artificial intelligence algorithms and/or rules, searches the knowledge base and integrated external systems, recognises the question and provides the user with a comprehensive answer.
Due to the functions they perform in a company, we can encounter the following division of text bots:
Information chatbots
They act as a knowledge base, answer frequently asked questions as well as communicate news
Trading chatbots
Streamlining e-commerce operations - a shopping assistant helps guide customers through the entire purchase process
Booking chatbots
enable the booking of different types of tickets or appointments (they are connected to the company's systems)
Chatbots for customer service, supporting contact centres
an online chatbot supports customer service departments by solving common problems or navigating the user through the site, in order to relieve the call centre consultant of the burden of answering the most frequently asked questions.
Recruitment chatbots/ virtual HR assistant
are able to receive application documents, carry out the initial interview, provide answers regarding hard HR. They also prove themselves during periodic knowledge tests on a given topic
Marketing chatbots
These include, for example, competition chatbots asking questions and accepting entries, conducting surveys and even... making customers' time more pleasant.
a chatbot for couriers,
a chatbot for couriers,
a chatbot for finance and insurance,
a chatbot for marketing,
a chatbot for HR/recruitment.
How does the chatbot work?
As we mentioned at the beginning, a bot is unequal to a bot – there are both basic solutions on the market that have nothing to do with machine learning and NLP and more advanced chat bots that rely on artificial intelligence (AI) and are capable of learning. Some of them will therefore only use a knowledge base and specific rules, answering in a schematic way precise and quite narrow questions. Artificial intelligence, on the other hand, gives bots the ability to ‘learn’, making them more versatile.
The foundation of the smart advisor is AI technologies:
the AI-based engine makes the robot conversational and is able to analyse and understand the question (along with the context),
auto-correction mechanism - the engine introduces simple linguistic transformations, i.e. improvement of spelling mistakes, use of Polish characters, improvement of typos, etc,
variation and lemmatisation - the chatbot is able to verify the words used in the query and reduce them to their basic form,
automatic response generation module - this works largely on the basis of a knowledge base containing factual data and also information that helps to create the impression of a normal conversation, e.g. "hello/thank you/question does not concern my area of expertise", etc.
finding in the message named elements and data supplementing the user's intention - e.g. the name of the product or type of service as well as related factors - feature (e.g. package weight), time, location , date, etc.
synonymisation - thanks to the synonym dictionary, words are also analysed for their synonyms,
the most common solution is a mixed one - a chat bot with AI that also uses defined rules.
Types of chatbots
Task-based chatbot
It works schematically and is predictable - it performs one main task or answers a closed group of questions based on rules - the information it conveys is repetitive, it does not go beyond a defined key. It is perfect for customer service in terms of so-called FAQs - answering frequently asked questions.
Conversation chatbot
Such a conversational robot doesn't just answer defined questions - it can be said to have a dialogue with users. Its operation is based on rules and also on machine learning, so that in addition to repetitive responses, it can learn and create a conversation similar to a human conversation. The chat bot is connected to powerful databases with significantly more resources than the database for the task bot.