Driving Business Value With Internal AI Assistants

By
Jasper Klimbie
October 2022
8 min
read

Driverless cars, robotic warehouses, self-driving trucks - sooner or later these are going to become an integral part of our life. AI has unlocked entirely new business models and shifted our ideas about how the world operates. An area that is still largely being ignored, however, is the organization itself.

You’re on the workfloor (remember that?) and you want to request a day off. What do you do? Trawl through pages of dense intranet content? Nope! You lean over and ask a coworker, or if you’ve been around longer, you don’t just ask any co-worker, but that one who ‘always knows everything’. Let’s call them Bo.

If Bo’s really awesome, they won’t just tell you how to do it, they’ll ask you three questions and get it sorted out straight away. That, in a nutshell, is the case for an internal chatbot. A single, conversational, source of truth and assistance for all those HR, IT and other internal processes. 

 

Content, content and more content

It’s well established that we must offer our customers efficient processes and service. Our hand is forced by the power of competition. But internally? Not so much. Often a new process doesn’t (fully) replace an old one and it’s simply added to the collection. 

So when you want to create a single point of contact and resolution, your team will need to work closely with all manner of owners and stakeholders. It is one the biggest hurdles in an internal chatbot project, but, over time, it leads to the streamlining of services and removal of deprecated processes. That’s a win, win. 

 

person holding black smartphone

Two types of approaches

For demonstrative purposes, let’s call our chatbot Bo, in honor of our heroic ‘know everything’ colleague. Starting out, there are two main approaches to go about it:

  1. Wide and shallow. Throwing out a wide net, trying to catch as much as we can. Also known as an FAQ bot. 
  2. Narrow focus. A deep and transactional bot that’s able to automate a (limited) number of manual tasks.

 

In the first approach, think of Bo as some kind of know-it-all, who can always direct you to the exact place where you find what you need. So, not some generic ‘catch all’ landing page, mind you. Because if you want your employees to adopt Bo as a colleague, and see them as a useful tool, you will need to prove your value to them.

In the second approach, think of Bo as an expert on a clearly defined subject who will take your query and help you sort it out, within the chat, in one end-to-end experience. Anyone who has ever worked in a corporate environment knows that a good part of your day is spent on uninspiring tasks: writing reports, filling out forms, logging your time, ordering stuff, categorizing documents or sending out orders. Today, most of these tasks could be partially or completely automated by a well-built AI Assistant.

Getting it to work might take more time and effort, also from the development side, but it can exponentially increase the productivity of team members or managers, when you’re able to successfully automate tasks that had to be done manually before.

 

Launching your first internal bot

As with any chatbot project, you would want to follow the three basic steps of the CDI Workflow. In the Requirements-phase you must focus your attention on finding the right platform and people to work on the project. Establish your approach. Decide what channels you want to deploy your chatbot on. For example, if your employees use Slack to chat, should your chatbot live on Slack and be able to hop in and out of channels? 

Once that’s all been ironed out, you’ll need to scope the first phase. Create user personas. A bot persona. Create happy flows of your first use cases. Test, implement, and optimize them. This will continue throughout the lifecycle of your internal bot as new domains and capabilities are added. 

 

Conversational AI with the human touch

Anyone who works in AI automation knows that there always is the looming risk of depersonalization. Some organizations might be tempted, at some point, to hire or fire people through AI. To reduce people to numbers and lists. To evaluate their performance by comparing it to hundreds of others doing the same job. 

At CDI Services we believe that conversational AI can be a powerful tool. But it is still a tool. It should be considered as a means to an end, and not the end in itself. We believe in building AI Assistants with a human touch. We should always ask ourselves the question: are the products and services that we’re developing are making our lives better… or worse?

Jasper Klimbie
Conversational AI Consultant
Share on:

Related articles