One of the challenges of last-generation connected devices was that they provided a lot of information, but no way to act on it. As algorithms improved, that information became insights that often suggested a particular course of action. And that is where most connected systems are today. Companies use these systems to manage cybersecurity, manufacturing plants, lighting, HVAC systems, and more.
But for many proponents of smart tech, the holy grail is having a system that detects a problem, or a source of more efficient operations, and then acts on that data. It’s what’s known as closed-loop automation, and these days, there is no one phrase that gets me more excited. Because ultimately, if we want to derive the most benefit from IoT and machine learning, we’re going to have to take humans out of the loop.
That is obviously a dicey proposition. There are jobs that would be lost, for one, but even more worrying is the danger of automating a failure and the potentially catastrophic consequences that could result. We can already see the challenges associated with closed-loop automation, such as when data centers go down after a networking error propagates through a system.
The key is building trust in the models used to develop the algorithm and understanding how the company offering the closed-loop automation sets the rules around the automation itself. A good example of a startup making this work is Verdigris Technologies. I met the Mountain View, Calif.-based company in 2016 and wrote about its energy-monitoring sensor and software.
At the time, the company counted Jabil, The Grand Hyatt, and others as among its customers. They attached the Verdigris sensors to each circuit in their respective electrical boxes so the sensors could track energy use at a granular level. From there, the data was sent to a building management system so customers could track how light bulbs, HVACs, and other energy-using systems were operating and if there was a chance to save any of that energy.
Thomas Chung, a co-founder of Verdigris, said that for many customers the system worked, but it frustrated them because it required a person to go turn off a light or lower the temperature in a particular area in order to achieve optimal savings. And so this week, the company introduced Adaptive Automation, which completely automates building management system responses to the Verdigris data.
The system ties back to external pricing data from utilities to ensure that buildings are not operating at times when rates might be higher or at loads that will trigger penalties. Obviously a building operator may not always be able to stay within lower rates, but Verdigris wants to make sure they can do so as often as possible.
Chung says customers should expect to see between one to two hundred actions taken by the system per month, or roughly a half a dozen or dozen actions per day. And surprisingly, getting customers to accept the closed-loop automation wasn’t as challenging as the Verdigris team expected it to be because those customers had already been using the system for a few years and trusted it. Chung also spent a lot of time building rules for when and how often the building management system could take suggested actions.
And lest you think that this isn’t an important issue for companies, know that most organizations have internal efforts or government-mandated efforts to reduce their energy use. So even if closed-loop automation wasn’t able to save companies money based on utility pricing, it would still be helping them meet internal or mandated goals.
In fact, energy optimization is one of the big trends I’m seeing among industrial and enterprise organizations as they add more sensors to their buildings and equipment. They now have the tools to understand how utilization, temperature, and outside factors can affect energy consumption. And as closed-loop automation becomes more accepted for more processes, we could see companies taking their efficiency plans to new levels.