With business rules, the answer to a question is pre-programmed in the system. Its advantage is that the system always returns the right answer to the question, changes can be done quickly, and it works from day 1 (as the system does not need any time to learn the link between a question and an answer). The disadvantage is that the system needs to know many synonyms so that the question can be recognised regardless of the way it was asked.
Machine ‘learning’ happens over time as the software becomes familiar with the context, characteristics and behaviour patterns of use. Unlike Business Rules, algorithms in the software analyse and interpret a logical course of action based on observations of past behaviour, rather than having to be specifically programmed. Machine learning software can analyse vast quantities of data and learn from it to deliver the best response it estimates to be true based on what it’s learnt so far. A simple example might be informing a customer that most people buy ‘product A’ when they have bought ‘product B’. And if bought at the same time, a discount is offered if bought today.
Machine learning is based on statistics and it works well with a massive amount of data. For example, if you want a machine to learn, that answer A belongs to question A, then this question needs to be asked at least 65,000 times. That’s why pure machine learning is not recommended to companies who would like to serve their customers quickly and efficiently.
The SmartStart is a great way to jumpstart an intelligent chatbot for your service department. In 8 weeks we help you to set up a basic database and a front-end so you will be able to go live. After Go Live, the DigitalCX dashboards will help you to rapidly improve the recognition of the chatbot assistant in one to two months.
Artificial Intelligence aims to imitate human characteristics and behaviours – such as interpreting and understanding, listening and interacting, recognising and personalising – particularly when it comes to customer engagement. Enabling machines to be perceptive and apply reason will ease repetitive tasks, such as the more script-based call centre functions but will need a great deal more knowledge to replace humans altogether for more complex tasks.
An API allows one software application to connect and interact with another software application across internal and external platforms, servers and databases. Third party developers can use or alter the software code to enable integration and collaboration with other software programmes. Acting as a ‘go-between’, APIs transmit information from one party to another to make a process work. A simple example is a comparison website, where one website application is used to send a message to other website applications and return the results using data that is pulled from external systems.
An API developer can choose to share their coding by making it publicly accessible via the Internet for other software developers to use. Programmers use existing APIs to develop and enhance functionality to suit their own requirements, using proven programming as their base code. The API originator can expand their user base while retaining intellectual property ownership.
More than a simple content management system, an enterprise content management system is content management strategy designed to store and capture structured and unstructured data from across an organisation’s processes. It is secure and connects with all APIs within the back office.
Dialogues in software are typically displayed using dialogue boxes which force specific action to be taken, data to be entered or criteria to be met before progress can be made and the process completed. Using dialogues for transactional purposes helps to accelerate the customer towards process completion as the conversation flow can be more predictable and, consequently, short and simple.
Algorithms are pre-defined instructions to follow to help get an answer to a problem or achieve what you’ve set out to do. A simplistic example might be making a cake. You may have all the ingredients, but what matters is the quantities of each and the order in which they are used. The recipe would act as an algorithm. In software terms, these can be complex mathematical rules used for problem-solving operations that calculate, process and analyse data and return the best possible result based on logical reasoning. Primarily used for speed, algorithms can also be intuitive, such as in ‘machine-learning’, which uses ‘best guess’ based on the information it has stored and the answer it predicts to be true. (see ‘Business Rules’ and ‘Machine Learning’)