Enterprise Chatbots: Your Workplace in the Age of AI, NLP, and Semantic Search
Chatbots are becoming ubiquitous. All around, Siri, Alexa, and Google Home are incorporating natural language conversations between humans and AI into everyday interactions. So how can chatbots bring this level of natural language understanding into the enterprise to improve information discovery and collaboration? This blog provides 5 key observations on the rise of enterprise chatbots and their use cases in the modern workplace.
What is an enterprise chatbot?
Powered by natural language processing (NLP), chatbots enable conversations between humans and computers in everyday business interactions. They bring deeper natural language understanding to not only enhance search but also provide an entirely new way for employees to interact with corporate data and work more productively.
For instance, an enterprise chatbot combining NLP, semantic search, and voice recognition can dialog with the user to acquire necessary information (dates, times, order details) or integrate with business systems to complete common office tasks quickly (reservations, orders).
Download our Guide to Enterprise Chatbots for an overview of chatbot use cases, technologies, and architectures.
What are some business use cases for chatbots?
Consider the following scenarios in which chatbots can help improve your employees’ productivity or customer service processes:
- Fill orders over the phone or via mobile apps
- Handle customer functions: order status, order cancellations, return instructions, tracking numbers, order modifications, account balance, etc.
- Act as personal digital assistants that help your employees do basic tasks: reserving conference rooms, registering mileage, recording expenses, etc.
- Provide automated support responses to customer inquiries
- Navigate through customer services, such as government or administrative services
Outside of the traditional office, organizations can also leverage chatbots for out-of-the-box uses such as interactive games, art installations, controlling equipment, etc.
When NOT to use enterprise chatbots?
Remember the times you saw Siri or Alexa fail to respond to certain questions or complete certain tasks? Similarly, while chatbots support many business use cases, you always need to consider if an enterprise chatbot is practical for what you intend to do. Here are some cases for which chatbots are not an ideal tool:
- Where language is unconstrained or widely varying: chemical names, diseases, people names, body parts, medical symptoms
- Where user queries and requests can cover a wide range of possibilities: cell phone problems, grocery store ordering, news search
- Highly technical domains where lay language is insufficiently accurate: car repairs, medical, academic, or engineering
- Where complex language is required: crime reports, general purpose searching, complex instructions
- Cases without discrete inputs (unless the options can be narrowed down for your customer base): house shopping, dating, deciding what’s for dinner, choosing a book
So what makes a good enterprise chatbot?
While your chatbot won’t need to pass the Turing test, creating a good business chatbot requires more work than it seems. Not only should the chatbot respond well to natural language queries, it also has to be able to handle rare cases or exceptions, such as:
- Providing alternatives to problems (e.g. an item is not in stock or a room is not available)
- Handling language differences
- Intelligently handling partial requests based on user’s history, past purchases, location, business unit, or the context of the request
- Normalizing requests (e.g. “Windsor” to “Conference Room 1-10”)
- Providing natural responses (e.g. “1pm” instead of “13:00”)
- Monitor what people say and modify the answers based on the actual user behavior
- Integrate with your mobile apps, business systems, and enterprise authentication/security measures
How to build an enterprise chatbot?
In general, an enterprise chatbot architecture integrates with your business data, search engine, NLP algorithms (download our full NLP technical tutorial or watch our on-demand NLP webinar), and in some cases, voice recognition, to provide answers or perform tasks.
From our experience, among various chatbot technologies available today, Amazon Lex (the AWS chatbot machine) and Amazon Lambda (an Amazon technology to execute small, quick-running programs in response to events) are solid platforms to start your project.
But as we tell our customers, before you begin building a chatbot, make sure you consider:
- Is a chatbot right for you?
- Are your current business systems ready to integrate with a chatbot?
- What technology stack/architecture would work for your business?
- Can you handle normalization of variations?
- Do you have a user interface device for your chatbot?
- What are the steps you have to take to build it?
Contact us to explore potential use cases and see an end-to-end demo of an enterprise chatbot built on Amazon Lex.
Download our Guide to Enterprise Chatbots to learn more about how chatbots can enhance productivity and insight discovery within your organization.
We’re very excited to announce that we’re now part of Accenture! Read the announcement here.