Many organizations have started developing chatbots, attracted by the allure of artificial intelligence’s promise of increased productivity, capacity and user satisfaction. As with any adoption of new technology, these initial experiences have been mixed in their results; implementation has been more cumbersome and complex than first imagined, and user experiences have been too robotic and off the mark. However, these lackluster results should not dissuade other organizations from jumping into the fray. In fact, we can now thank those brave early adopters for paving the way to understanding what works and what doesn’t.

Chatbots are a compelling technology to support a corporate learning and development (L&D) program, because they address the fundamental challenge of learner engagement. Chatbots are designed around a conversational user experience, which means that engaging with a chatbot can be as frictionless as engaging with a colleague or a friend. If a chatbot is programmed to be “smart” about a user’s learning journey, it can easily exchange messages with that user to remind, nudge, teach and hold accountable — all activities that will maximize the impact and return on investment (ROI) of your training program’s investment.

How does one go about designing a chatbot? And just how smart can a chatbot be? Virtually every major technology company is investing significant resources toward improving its artificial intelligence (AI) capabilities. Still, we are a long way away from AI bots that are as conversant as, say, Ironman’s J.A.R.V.I.S.

The primary design sensibility when building your first learning chatbot is to be disciplined in defining the scope of what the chatbot does (and doesn’t do). As such, a useful paradigm is to treat your chatbot as a member of your L&D team and give it a specific assignment, just as you would a new hire. Here are some attributes to give your chatbot team member:

Job Title

Do you remember when Apple introduced Siri to the world, and people started asking Siri all sorts of random questions like, “Siri, where do I hide a dead body?” (try this one if you haven’t already!) or, “Siri, what is the purpose of life?”. People have a tendency to test a bot’s depth of knowledge, and it is impossible for any bot to deliver a satisfactory answer to all inquiries.

To help mitigate this challenge, give your chatbot a name and a title. The more specific the title, the more clearly you communicate what the chatbot is capable of. Here are some examples:

  • “Hi, I’m NEO, your new employee onboarding chatbot!”
  • “Greetings! I’m ACE, and my job is to help you remember what you learned in sales training.”
  • “Hello. My name is Sam, and you can ask me questions about our safety and compliance policies.”

Rules of Engagement

Some executives worry that people don’t want to deal with bots but would rather interact with human beings. On the contrary, most people want their issues and questions resolved quickly and correctly, and they don’t mind automation if a bot can help them quickly. They just need to know how to use it.

In addition to clearly naming your chatbot and giving it a “job title,” provide clear instructions to users about how to use the chatbot:

  • “Hi, I’m NEO, your new employee onboarding chatbot! I’ll be sending you daily reminders and tips as you start your new job. If you have a question for me, just type ‘menu’ at any time.”
  • “Greetings! I’m ACE, and my job is to help you remember what you learned in sales training. I do this by sending you a daily quiz question. You get to choose the time of day to receive this quiz. What time of day works best for you?”
  • “Hello. My name is Sam, and you can ask me questions about our safety and compliance policies. Just enter your question in the text box.”

Knowledge Base

After clearly giving your chatbot a title and a defined rules of engagement, you can feel confident about limiting the scope of content your chatbot needs to be aware of. In fact, your chatbot can be an effective concierge by pointing people to existing content resources. This approach saves you a lot of content creation time while encouraging people to consume existing resources that, all too often, go under used. You can easily program your chatbot to return search results and pass along links, images, videos and PDFs.

Here’s an example:

Chatbot: Hi, I’m NEO, your new employee onboarding chatbot! I’ll be sending you daily reminders and tips as you start your new job. If you have a question for me, just type ‘menu’ at any time.

User: Menu.

Chatbot: Select from the following: HR forms, benefits, tips, coaching, training courses.

User: HR forms.

Chatbot: OK. Here is a link to our intranet with a list of all our forms.

It is also useful to program your chatbot with answers to the most commonly asked questions — typically a manageable amount of content to create and maintain.

Chatbot Limitations

No matter how well you design your chatbot, it is impossible to guarantee that it will be to answer every question or deliver the right message every time — but the same is true of any human on your L&D team.

What do well-trained human team members do when they are unable to help a colleague? They work to find the right answer, or they point their colleague to the appropriate source. You can program your chatbot to politely do the same thing.

Example 1:

User: How do I give feedback to my manager without threatening our working relationship?

Chatbot: That’s a great question. I’m really not equipped to answer it. We do have a list of coaches who help with these types of situations. Would you like to be put in touch with one?

Example 2:

Chatbot: Hi there. This is your friendly Friday afternoon reminder to review the new product training you took last month. Here is a link to a PDF with summary info:

User: Thanks. Is this link available in a Word format?

Chatbot: Oops, I didn’t understand your question. I’m going to forward your question to my support team to get you an answer.

By treating your chatbot as a team member with these attributes, you can reap the rewards of the power of chatbot technology. Unlike human team members, a chatbot never has to sleep, can talk to thousands of people at once, can speak multiple languages and never forgets a conversation.

Chatbots can be a game-changer in helping your L&D team influence your organization by freeing up your time to be more strategic and invest in the relationships that will support your initiatives. So, don’t be intimidated by this new innovation. Pick up your pen, and start designing your latest team hire: a chatbot.