What is an Enterprise Chatbot?
Enterprise Chatbots are using most of the technology regular chatbots use to create use cases tailored to a given organization and with the targeted employees in mind. Examples would include:
- Chatbot for IT Helpdesk/Servicedesk issues that suggest solutions based on the problem description and open a ticket if the problem couldn't be resolved
- HR related chatbots for asking about policies, allowances, workplace perks, etc.
- Business intelligence chatbots that could look up reports, facts, figures and maybe even send visualizations in the form of charts, gauges and tables
- Intranet chatbots that would tie into multiple searches and can help narrow down search results with Natural Language Processing (NLP) to capture file types, times, authors or storage locations
- Subject matter expert chatbots that could be an intermediary between lengthy knowledge bases and employees trying to answer ad-hoc questions for either themselves, or even customers they are speaking with in a shop or over the phone
Enterprise chatbot challenges
A major obstacle for creating an enterprise chatbot with a high adoption rate is data binding. Admins will find it easy to set up text responses and teach the chatbot the necessary Natural Language understanding to safely match a question to an answer.
But often times the answer will be 'canned' and merely contain a hyperlink to a knowledge base article somewhere on the Intranet. Or it gives the user a lengthier, non-specific response they need to read through to pick what is applicable to them. That places nearly the same cognitive load on the user as just going to the Intranet search or Enterprise search portal directly. Thus often chatbots end up not leaving the 'experimentation' stage, as they're not sufficiently better than what was before – thus failing to 'retrain' user's habits.
In order to give enterprise chatbots the capacity to give the user a personalized, direct answer to their question, the chatbots needs to be integrated with all existing business applications from both the cloud and on-premise sources. This requires the chatbot platform to offer API integrations and a 'designer' or 'configurator' that lets the admins pick out which part of an API response would be used in the answer given to the user.
The goal would be for the chatbot not to say "You need to click here to log into the portal to find out how much leave allowance you have", but instead just respond "You have 6 more days this year. Want to request more?"