RigD Platform for Automated Response Part 3 – Natural Language and Machine Learning

• RigD Platform for Automated Response Part 1 – Use Cases and Response Steps
• RigD Platform for Automated Response Part 2 – Core Capabilities for ChatOps

Even thought we are discussing Machine Learning, let’s take a look at what an enterprise does with Bots “in the wild”. I was chatting with a colleague of mine and he mentioned that at his company that they have Bots coming out the woodwork. These Bots were created by everyone, each was different. These Bots also were typically set up to do just a few things.

While we can all see that having a structure where Bot capabilities where like the modern microservices architecture, one can easily see how this gets out of control. Many of these Bots have no owner anymore as that individual has moved on. They do not have a similar taxonomy of naming or even how your interact with it. Let’s not even get into looking at this from the code level. As an Manager of the SRE team, you are responsible for automation, but do you want to let the wild west rule in your organization. After all, everyone is highly capable and nothing breeds innovation like a real life problem and a team focus on automation.

But should you let your org build Bots without dare I say the “governance” that you apply to your business logic?

Here are a few resources we have written on the subject:

Bot Behavior Fundamentals

Universal Natural Language User Interfaces and Seamless Workflow Automation

Bots: ‘Like a CLI on Steroids’

Natural Language Understanding – The Art of the Possible

The Natural Language Understand (NLU) for Bots for IT Professionals are very different than say what Siri or Cortana has to deal with. For one thing those Bots are in the business of entertaining us and helping us find out information from any number of sources. Also they help us do things within the scope of the device it is running on. Above all Bots for IT professionals have to listen to the arcane talk of IT and DevOps and span the scope across many tools with APIs. The RigD Bot has to be able to “drop down” into CLI type commands as that is needed occasionally

We have a specialized Natural Language Processing engine that’s designed for technical interactions. Therefore our NLU is continually retrained using real values for technical content. This model retraining is critical in helping our Bot infer the right command from a conversation thread. The RigD NLU makes use of these capabilities:

• Activity Observation
• Pattern Identification
• Action Prediction
• Suggestions
• Predication Validation
• Building Automation

The way this all occurs is through contextual models based on an individuals or teams unique content. We offer faster & more accurate execution of machine learning augmented activity. Below is a picture of RigD providing context aware suggestions for input.

RigD also allows you to alias commands to make it easy for the team to quickly execute tasks and activities.

RigD can also identify potential automation flows based upon repeated observation. Suggesting automation take a huge burden off teams to conceive and write it on their own and running it with user defined text enables better recall.

Obviously, your team can create a purposeful Bot to do a specific task, but will your team build, maintain and support a wide variety of Natural Language Understanding and Machine Learning capabilities? Even organizations with dedicated tools groups have challenges keeping up with the latest in these spaces. The RigD platform is the answer to this challenge. Give our Bot a test run here.