Greg, you seem to excel at designing cutting-edge software. First Enviance, and now ehsAI. What in your background has contributed to your talent for this? What do you love about this work?
Beginning my career, I worked at the US EPA Air HQ office and was part of the teams developing New Source Performance Standards. I then spent 5 years as an air quality inspector and unit leader at the Louisiana DEQ, and two more years as Assistant Secretary there, in charge of the Air Quality and Radiation Protection Divisions. I learned compliance from the perspective of the regulatory agency. I then spent 13 years as a compliance consultant to industry, helping industry respond to the rapidly changing laws and rules. It was during those years that I began to see the “data explosion” that was resulting from the laws that were being passed. I realized that there had to be a better way of managing compliance, one that more efficiently responded to massive record keeping and reporting requirements. It was that recognition (and improvements in internet technology) that allowed Environmental Management Information Systems, like Enviance, to take root.
I love the challenge of creative problem solving. For me, experiencing the problem on a day to day level gives me the incentive to create a new solution. At Enviance, I experienced the difficult job of manually extracting obligations from permits and rules. I knew in my bones there had to be a better way. It was that pain and the improvements in technology (Artificial Intelligence) that led to the opportunity to partner with Margery to create ehsAI.
ehsAI is pretty disruptive, how do you see ehsAI changing the compliance management services industry?
ehsAI will free up valuable human resources to make improvements at their industrial and business facilities. Let’s face it, manually extracting obligations from permits and rules is tedious and painful. As ehsAI is more broadly used, it will allow those who were previously doing that work to do higher value things.
What is the most unique thing about ehsAI that makes it stand out?
The ability to combine multiple permit conditions or citations in a rule into a single requirement. This feature truly mimics the way the human brain reads a set of text and gives the proper content and context for our customers. I am also very proud of the high degree of detail and granularity in ehsAI. We review each sentence of the entire text.
We are all new to AI and ML, what can you tell us about working in this experimental medium now that ehsAI is out and commercial?
My previous designs were all rules based. So, designing a feature or function was primarily about describing the desired outcomes for a given set of data conditions. Understanding the problem and good technical writing were mostly what was needed – the rest was up to my colleagues writing the code. Machine learning is a more powerful tool and, for me, a more fluid approach. With ML, I do not have to solve a problem head on, in a single step. I can break the problem down into smaller components and use different ML steps to accomplish the end goal. The effect of this highly flexible approach is the ability to solve tougher problems.
Since AI is so new, what do think are the biggest challenges?
Frankly, it is the incredible variability in the way regulatory agencies write permits and rules. That variability means we need very large amounts of training data. For example, we do not classify actions required of the regulatory agency as official Requirements because they do not incur an obligation on our customers. Yet the syntax and wording of such citations can be very similar to those citations that do incur obligations the regulated entity. It is not like other forms of AI, in which training data can be generated by any individual saying if an image contains a dog or a cat.
How are customers receiving ehsAI and using it? Any surprises?
The reception has been fantastic. I especially love the reaction from those who have spent significant time deconstructing rules and permits when they first see a complex 70- or 80-page document get processed in a couple of minutes. Regarding surprises, I have been surprised at how many different adaptations our customers have asked us to consider. ehsAI fits within many larger compliance workflows, especially for the very largest companies. I can definitely see custom adaptations in which we build outward from ehsAI to solve bigger problems.
There is a lot of opportunity here, where would you like to take ehsAI?
For starters, I see an end-to-end solution beginning with Applicability Determination all the way through to Evergreening. Beyond these natural extensions of ehsAI, I believe we can unify the compliance process across many disciplines like Environment, Health, Safety, Transportation, Food Safety, and Supply Chain. The better we get at addressing the variability inherent in regulations and permits, the more applicable we become to all compliance processes.
You must be super busy, but what do you do on your downtime?
I live in New Orleans and really enjoy the music and cuisine here. I try to take my sweetheart (Laurie, my wife) out dancing a couple of time a month. I also like to cook, especially the things that I learned at home as a boy. Dishes like Redfish Courtbouillon, Crawfish Bisque, Oyster Dressing, and Shrimip Etouffee are my favorites. I also like to catch the fish I cook, when I have time. I love fishing in the marsh on a cool, sunny day!