There are at least two use cases.
The first one is an internal concierge, when robots help out with ordering stationary, arrange travel or help onboarding new employees. The more cloud solutions are in use by an enterprise, the easier the adoption of the tools, which simply interconnect many such systems (“integrations”). The key is of course public APIs of those systems. Otherwise it takes skill and time of internal development forces and certainly limits the global adoption of such tools as chatbots.
The second use case is customer support and pre-sale automation. Take a look at Intercom’s solution and trends. The main barrier here is the way the big data is stored and the technology available for a regular company to process and label it.
The reason is that the ML tools are still not that easy to use. We need it at the level that an ordinary IT person could use it like he or she uses bash scripts nowadays.
Another issue is regulation around the data (such as GDPR) scares many enterprises and they prefer to stick to the known solutions. It is hard to persuade such a company to move on with the cloud chatbot solution. And since many of the enterprises still don’t have ML as a core expertise, they can implement a rule-based chatbots at best. More often, this leads to poor performance and bad customer experience. Disappointed, such companies shelve such projects for some time, maybe even years.
However, with the great adoption of voice assistants we are bullish on conversational UIs including chatbots!