What is decision automation?

It’s new and yet old. It bridges data science and business operations. And it’s an emerging piece of critical infrastructure for realizing a more efficient, responsive, and predictable world.

Today's businesses — big and small — have to make the best decisions in the time they have, and in a scalable way in order to compete. Just like search, transaction processing, and data management, decision automation requires the right technology. 

At Nextmv, we’re building a developer platform for automating and optimizing the decisions that drive every operation — from vehicle routing to workforce scheduling and beyond. Put another way: Nextmv is a decision automation platform. But what is decision automation?

Decision automation is infrastructure for modeling, implementing, testing, and managing decisions as code.  

Is this new? Well no and also yes.

Paradigms and frameworks around decisions are not new. Optimization, simulation, heuristics, business process management, Six Sigma, robotic process automation, and artificial intelligence (AI) all impact business decisions. 

"Methods for solving optimization problems are equally numerous and provide a large reservoir or problem-solving technology. In fact, there is such a variety of methods that it is difficult to take full advantage of them. They are described in different technical languages and are implemented in different software packages. Many are not implemented at all." -John Hooker, Integrated Methods for Optimization

Traditionally, in industries like defense and manufacturing, engineers with backgrounds in industrial engineering, systems engineering, or operations research implemented these paradigms or applied these frameworks through on premise software solutions and bespoke applications. They were mathematically intensive (think of missile launches or a perfectly orchestrated assembly line), required digitized data feeds, and used extensive compute power (the accuracy of a missile intercept is pretty important). 

Cheap cloud computing made it so that every company could utilize these decision making paradigms by sparking multiple waves of data trends: 

  1. Digitization: we can observe information about our system (drivers, operators, consumers, etc.)
  2. Data storage: We have ways to store and access that information in real or near real time.
  3. Business intelligence: we understand what happened in our system and how it impacted our business. 
  4. Data science: we can predict what may happen in our system (forecasts, consumer preferences, etc.)

Despite all the data access, they still struggle to bridge the gap between insights (what is going on?) and business operations (what action should we take?). 

This prompted a fifth wave of decision intelligence. This emerging discipline turns information (data and predictions) into better decisions at any scale by applying technology, paradigms, and frameworks (everything from applied statistics to ML to optimization). Decision intelligence gives legacy disciplines of systems/industrial engineering and ops research a common home alongside more recent additions like data engineering and data science. 

With decision intelligence, companies can design interconnected human and technology systems to tackle real-world problems: How should I schedule my workforce to meet my demand? What’s the best way to allocate my inventory into subscription boxes? How can I pack my delivery truck to save my drivers the most time? Decisions and decision flows cut across all aspects of the operation.  

Tired of the word decision yet? We’ve been swimming in the soup of decision terminology for a while. And we aren’t the only ones. 

IBM states that “decision management is the process of using machine learning and business rules to automate decision making.” 

In Gartner’s view, “Decision intelligence is a practical discipline used to improve decision making by explicitly understanding and engineering how decisions are made and how outcomes are evaluated, managed and improved by feedback."

Google even added a Chief Decision Scientist in 2018 “to bridge departments that usually keep to themselves, all the way from research to the teams that apply algorithms to business functions.”

So why add decision automation to the mix? In order for decision intelligence to be successful, there needs to be infrastructure (decision automation platforms) and engineers to systematically realize it. Where decision intelligence is a discipline, decision automation codifies its principles and exposes decision making technologies like optimization in with a set of standardized primitives. This gives engineers the infrastructure to implement and manage powerful decision making components across a variety of problems.

The decomposition of data, analytics, and AI platforms into smaller reusable components allows organizations to rapidly compose and recompose transparent decision flows. The resulting more continuous, connected and contextual decision intelligence solutions enable more dynamic, yet optimized supply chains; or more scalable, yet personalized customer interactions. - Gartner

While automating decisions isn’t exactly new, making it developer friendly is. Our decision automation platform blends hard core math with new software primitives to democratize one pillar of the decision intelligence realm. In the same way Twilio gave engineers new primitives for messaging, we give engineers the primitives to impact business decisions through code — from how we get our groceries to how we care for the environment. From where we stand, decision automation makes the world more efficient, responsive, and predictable.

We’re a little too pragmatic for hype, but here’s to adding more flavor to the “decision” soup. 

Video by:
No items found.