What is the difference between the three options?
The easiest way to create a chatbot is surely via one of the many chatbot providers to click static dialogues together via a graphical interface. For example, to find the opening hours or address of a business. However, you will reach the limits as soon as external data sources from business systems need to be integrated into a conversation. The same applies for intelligent user conversations using AI technologies.
Alternatively, it is advisable to develop a chatbots yourself. In principle, these degrees of freedom to integrate any business systems, any messengers and AI systems is enormously. However, this also applies to the costs of implementation and maintenance of chatbots, which quickly make integration uneconomical.
Usage of a Chatbot Platform
Companies can profit to unify the benefits of both options. Therefore it makes sense to use a chatbot platform that allows flexible integration of business systems and AI technologies, without having to redevelop all the essential components such as interfaces to messengers and AI systems. A chatbot platform is similar to a LEGO kit (“bottom-up strategy”): The basic functionalities (building blocks) are ready to implement an application scenario (model).
Which option is interesting for who?
The selection of an option is driven from the respective application scenario and the individual requirements derived to the chatbot. For example, how intelligent the chatbot should be, but also requirements for messenger services and business system integrations. Basically, the competence of a chatbot should be high enough to ensure a positive user experience. If this is not the case, there is a high probability that the user will not use the chatbot service a second time.
The following table provides an overview of fundamental selection criteria.
|“Click-together” Chatbot||Self-developed Chatbot||Usage of a Chatbot-Platform|
|Realization of the chatbot use case|
|Do I like only static content as a chatbot answer?||Mostly only static content is possible||Static and dynamically generated content can be realized||Static and dynamically generated content can be realized|
|Do you want to use information from external data sources and business systems? (e.g., ERP-, CRM-Systems)||X||X|
|Should the chatbot act intelligently? (Integration of AI technologies)||Mostly not possible||Freely selectable||Mostly freely selectable|
|Are programming skills mandatory?||Not mandatory||Always mandatory||Partially required if own use cases need to be realized|
|Should the chatbot be extensible with libraries?||No||Yes||Yes|
|Should other instant messengers be used or integrated?||Defined by the provider||Freely implementable||Defined by the provider, but usually expandable|
|Should the chatbot interact with the company’s IT infrastructure? (e.g., user identification via Active Directory or LDAP, loading of user data)||Not possible||Freely implementable||Prefabricated modules freely selectable and usually expandable|
|Entrepreneurial and economic aspects|
|Do you have high data protection requirements in your company?||Low||Very high||High (PrivateCloud) to very high (onPremise)|
|What are the costs for the realization?||Low||High||Low to medium|
|How much time is needed to realize a chatbot?||Due to limited chatbot intelligence without external data source connection: <1 day||With chatbot intelligence and external data source connection: ~ 3 – 6 months
(depends on data source)
|With chatbot intelligence and external data source connection:
~ 3 days to 2 weeks
(depends on data source)
|Would you like to have privacy under your own control?||Public Could: Defined by provider||Fully under your control (exception for public instant messengers)||
Public Cloud: Defined by provider
Private Cloud: Mostly under your own control
onPremise: Fully under your control (exception for public instant messengers)
|Should the location of the installation / provider be freely selectable?||Defined by provider||Freely selectable|| Private Cloud and onPremise:
|Which type of hosting is preferred? (Cloud / onPremise)||Only Cloud||Cloud and onPremise||Cloud and onPremise|
Is any trend apparent for corporate communication via messengers and chatbots?
The penetration to use instant messenger and chatbots is covering all business areas: from human resources, marketing, to internal IT support. However, all areas have one thing in common: all chatbots used in various business units must offer users added value in order to establish themselves permanently in the company.
Especially in areas requiring fast communication, the instant messengers will become more prevalent and thus opening up enormous potential for chatbots. Slack is a good example that allows users to communicate through channels, while chatbots automatically provide valuable content (such as sentiment analysis and competitor information).
What is important at this point when using instant messengers (and chatbots) for communication?
Many chatbot prototypes do not go beyond a technological demonstration, as they usually lack a positive user experience and real value. A chatbot without the integration of business systems inevitably leads to a limited technological implementation with low added value. Chatbots can only increase added value, when external data sources are connected which extend the overall competence. In other words, it can also be said that chatbots only represents simple puppets without the expertise of external data sources.