For many organizations, the recent rush to the cloud has also given pause to strategic considerations around what best to leave in house. Clean Harbors is one such entity taking a hybrid approach to its digital transformation to ensure high availability of key systems while still gaining access to advanced AI and automation capabilities where necessary in the cloud.
Roughly seven years ago, the hazardous waste cleanup giant took baby steps into the digital world by embracing Salesforce CRM. Two years later, the Microsoft shop migrated to Office 365.
Soon thereafter Clean Harbors took a big leap to Microsoft Azure’s AI Cognitive Services and Azure Machine Learning Platforms to gain valuable insights into its operations, adding robotic process automation (RPA) platforms from UiPath and Automation Anywhere to automate business processes as well.
But the Norwell, Mass.-based company — which played a key role eliminating hazardous waste during the COVID pandemic and in the aftermath of 9-11, Anthrax attacks, Hurricane Katrina, and the massive BP oil spill — continues to rely heavily on its on-premises SQL-based Waste Information Network (WIN) and is in no hurry to migrate everything to the cloud.
“Our journey into the digital world started when we had the desire to start doing AI and RPA. Our core areas, CRM and ERP, are in the cloud,” says CIO Sharon Gabriel, noting the company recently migrated to Oracle Fusion HR in the cloud as well and expects to migrate its Oracle Finance and Supply Chain modules and its UiPath and Automation Anywhere RPA systems to Azure within the next two years.
“Our strategy in taking a hybrid approach has provided the agility we need to do advanced services in the cloud as we go through our digital transformation,” says Gabriel, who joined the company in 2001 and was promoted to executive vice president and CIO of Clean Harbors in 2018.
Much of Clean Harbors’ customer data remains on premises in roughly 50 SQL databases that are integrated with AI ML models through APIs and with RPA bots — a hybrid approach increasingly familiar to IT leaders looking to take strategic advantage of cloud capabilities while also retaining control over key systems and data.
Cautious, methodical approach to the cloud
Moving aggressively to the cloud is not a high priority for Clean Harbors. Early adoption of technology in the past has led to problems — and there’s 42 years’ worth of data stored in on-site databases, its CIO says.
“We’re not early adopters because sometimes you get burned, and we got burned, so we’re going to make sure that the cloud is tried and true before we just jump on the bandwagon,” says Gabriel. “And it’s a little bit difficult for a person in my shoes to hear about all of the amazing technologies that companies are doing in the cloud and what an enabler it is to scale and move quickly and not jump on the bandwagon.”
Because of this, Clean Harbors’ CIO and the company’s Cloud Steering Committee are committed to a hybrid model.
“We will continue to utilize the advanced capabilities of the cloud for customer portals, AI, and soon RPA, which deliver big efficiencies and a huge return on investments with lower risk,” Gabriel says.
“Our company has grown through several major acquisitions, and it is imperative we present one face to our customers with the diversity of our brands,” says Gabriel. “With customer-facing applications in the cloud, we can present this posture, adapt, and change direction rapidly. We can configure and scale to have the right solution quickly in the areas where we need it to support our customers.”
But when the company was founded more than four decades ago, there were no commercial off-the-shelf systems for managing hazardous waste, so Clean Harbors developed its WIN quoting and billing system, which it still runs on premises today — a key system that represents “everything we do to run our business,” Gabriel says.
“It’s all custom dot.net and that data all resides in our Power BI solution so we can have robust dashboarding and reports. We have a snapshot environment, a repository of everything, all the time, constantly being refreshed,” she says.
The company’s low tolerance for availability risk has been a key factor in the decision to keep WIN in house, Gabriel says. “As we take advantage of the benefits of the cloud this need for high system availability helps us determine where and how we get there, … and we leverage the committee to provide governance and ensure we take availability and other security risks into account.”
Clean Harbors is not alone in its hybrid approach to the cloud. According to Gartner Group, enterprise IT spending on public cloud computing in leading market segments will not overtake spending on traditional IT applications until 2025.
Although the pandemic accelerated uptake of the cloud, the “cloud shift” in total IT spending isn’t expected to reach 51% until 2025, compared to 41% for cloud spend in 2022, Gartner’s research claims.
Embracing the cloud where its impact is greatest
Still, Clean Harbors has made a big investment in Microsoft Azure, in particular to leverage its AI capabilities. The company’s 400 IT staff — located at its Norwell, Mass., headquarters and in Hyderabad, India — has created hundreds of AI models, including automated billing models, customer churn models, and service frequency models, Gabriel says, adding that Clean Harbors may have as many as 15 AI projects under way at any given time.
The crown jewel in Clean Harbors’ AI portfolio, though, is its industry-specific Profile Model, which defines waste codes as required by the EPA and Department of Transportation, she says. First deployed in 2020, the Profile Model is core to Clean Harbors’ sustainability business, defining what type of waste its customers want to eliminate and helping make decisions about how and where it will be disposed. Developers integrate Profile with WIN using APIs.
“Our customers fill out seven pages electronically of all the specific details of their waste — the substance, the physical characteristics, the composition and constituents, and our chemical product group will look at the data and assign a waste class code,” Gabriel says, adding that developers created three different Profile AI models to ensure they accurately define the waste class code.
“And based on that waste class code, we know how to treat it and we have another system where we run low-cost routing, so we send it to the disposal facility,” she adds.
It’s critical that the inputs are accurate — or there could be an explosion at one of the company’s 650 facilities nationally, or inside a truck on the road, Gabriel says.
As a result, all AI models — especially the Profile Model — are continuously tweaked and updated to ensure customers get the most efficient quotes and pricing for waste removal.
“Not a lot of people deal with hazardous waste and not too many companies can say they built an AI model around hazardous waste codes,” Gabriel says.
Clean Harbors’ hybrid cloud approach is likely to remain the model for many enterprises that want to exploit cloud services while maintaining core systems on premises, one analyst says.
“There is little doubt that cloud has made technologies like machine learning and artificial intelligence more accessible, and companies want to take advantage of those to drive innovation. But that doesn’t mean they need to go all-in on the cloud,” says Dave McCarthy, research vice president of cloud and edge infrastructure services at IDC.
“At one point in time, there was a belief anything and everything would migrate to the public cloud. However, nothing is ever that simple,” McCarthy adds. “There are many situations when workloads need to remain on-premises, whether due to performance, cost, or corporate governance policies. That has created renewed interest in hybrid cloud architectures, where it is possible to mix both traditional IT and cloud environments.”