Professional Services

Combination Plays: Automation + AI or ML

Neil Ward-Dutton of IDC talks about three of the ways AI might augment automation: Conversation, Document Understanding, and Predictions. Watch the video.

Neil Ward-Dutton of IDC talks about three of the ways AI might augment automation: Conversation, Document Understanding, and Predictions. His latest video series is a great primer on a number of key business issues around Intelligent Automation solutions!

One example of a conversational system: a chatbot with more language capacity to allow a small team to service a wider geographic and linguistic surface area. I always thought chatbots would take off in the same way that automated receptionists have taken off, but the progress has been a little slower.

Document Understanding - Neil is right that historically OCR was considered too brittle.? Modern versions of this software not only read the documents but also extract data elements: an address; an introduction; a signature.? Not to mention that text can then be fed to sentiment analysis as well. These solutions tend to come to market paired with RPA (Robotic Process Automation) vendors. We recently wrote about these on our blog under the heading "Intelligent Document Processing"

Predictions of next best action. These models can be done with normal statistics, but it is less practical to do the work this way versus using a machine learning approach that can be flexible to changes over time. As Neil points out, these come to market more from process vendors.

What AI and ML techniques really benefit from when combined with a chatbot, an RPA bot, or a process (and a person) is agency. In other words, if AI or ML technique is the brain, observing the context and inferring the right thing to do, it then needs hands (or feet) to do the right thing. It doesn't just happen.

Knowing the next best action without knowing when and where to take that action is less than useless. Process gives that next action context for the when, the where, and the what next.

Knowing the meaning of a document and the data entities doesn't mean a thing if you don't have an RPA bot to punch that information into the right system, or to pass the information along to the right people to look at.

Knowing what someone asked in a chat window doesn't help at all if you don't have a way to convert that into a meaningful response to publish to them.

Whenever you're thinking about artificial intelligence, think about two things:

  1. First, think of it as augmenting intelligence, and then ask "What is it augmenting?"
  2. Second, after the AI gives you an answer, think "And then what?" as in, and then what should happen next? And think about how to make that happen, mechanically.

Answering those questions will help steer you straight. Or, you can always hire BP3 to help you work through it - after all, this is what we do!

Similar posts

Want to stay up to date with BP3's insights?

Subscribe to our newsletter