YOUNGSTOWN, Ohio – Companies and their technology service partners continue to evaluate how best to use artificial intelligence, area technology providers report.
Implementation of the emerging technology is “a little slapdash at best,” according to Robert Merva, owner and CEO of Avrem Technologies LLC in Canfield.
“There is no threshold of size or revenue or employee that makes AI valuable, because its primary use today is within automating documentation, contract review, license review, investigating and researching topics,” Paul Hugenberg, managing member at Pelican3 Consulting in Poland, says. “Everybody can do that, whether you’re one person or 1,000 people big.”
How AI is used largely depends on the industry, Ralph Blanco, president and CEO of ECMSI in Struthers, says. Customers, including those in manufacturing, often use AI in sales and marketing to “be creative.” In the medical field, staff are so busy with “day-to-day stuff” that having applications to help with patient diagnosis by looking at data to provide recommendations provides a competitive advantage.
“People are still trying to figure out how to use it, how to take advantage of it, what platforms to use, what their strengths and weaknesses are,” Merva says. “I’m not seeing widespread adoption. I am getting a lot of questions as to how AI can help, and we’re trying to help clients navigate those discussions, but I’m not seeing the kind of implementation that you might expect – not yet, anyway.”
Customers aren’t calling to ask how they can implement more AI, according to Blanco.
“We’re bringing it to them to help them become more efficient,” he says. “When you get into most industries, customers are so busy doing their day-to-day work, unless they’re IT providers helping them with that kind of stuff, they’re not really looking at AI as far as their business.”
Types of Uses
Customers are implementing AI in their operations in a couple different ways, Hugenberg says. Most people are implementing AI in “kind of the ChatGPT sense,” to draft documents, write policies, evaluate contracts and send emails. It also is being used to create “agents,” which automate repetitive tasks, such as notifying a salesperson that a file has arrived.
“Anybody who does a lot of repetitive tasks is trying to automate things that weren’t previously automated and use AI to kind of speed up that process a little bit.” Merva says. “We’re seeing a lot of insurance-related things. We’re seeing a lot of things related to sales and marketing, copy generation, kind of the more obvious areas that you would expect AI to kind of take a lot of the burden off people.”
In manufacturing, AI is being used in the Internet of Things architecture, Hugenberg says. Applications running on IoT devices are operating proactively, looking for when devices might break down, how much capacity a line is producing and whether pounds per line is decreasing because of operational efficiencies or lack of maintenance, for example.
“The plant manager or the warehouse manager is now getting proactive information that says line two will likely suffer a problem in the next 90 days because of these issues, and here’s what to fix,” he says.
Clients report efficiencies from use of AI in applications such as note taking and meeting summaries, “eliminating the whole workflow that they used to have around those things and automating that,” Merva reports. Many of the gains are coming from sales and marketing operations, which are able to pump out more content than they were previously able to before,” he continues.
“We see a lot of success when people basically ask AI to ask them questions, and it kind of acts as a sounding board,” he says. “That’s been very helpful and beneficial to people.”
Everybody uses AI a little bit differently, Blanco says. “We use it internally to help with automation and that stuff for our customers – onboarding, offboarding … how do we reduce the amount of tasks we have to do using AI to supplement that,” he remarks.
ECMSI is “using it more internally,” to help make its technicians more efficient, by providing potential solutions before they have to do anything, Dave Galioto, ECMSI’s chief information officer, says. It also is used to set priorities on tickets and provide alerts about unhappy customers.
“It also helps us find if there’s reoccurring issues that are happening, to help us make our clients more efficient,’ Blanco says. “We’re really trying to help their operations by using AI internally, from an IT infrastructure standpoint, to help make them more efficient, help reduce downtime and those types of things with our internal AI.”
Regardless of size, any business that relies on technology is using AI, whether it’s intentional or just a byproduct of the software they are using, Hugenberg says. “If you’re using Excel, you’re using AI. If you pick up your iPhone, you’re using AI. If you drive your Tesla, you’re using AI,” he says.
Decision-Making
The businesses that are seeing success with AI aren’t just using it for documentation and sending emails but are recognizing they need to automate processes to help them make better decisions, Hugenberg says. For example, to make hiring decisions, they create an AI routine or agent that automatically scores job candidates against an established benchmark based on personality reviews and pre-hiring checklists.
“So, now I’ve cut out a major part of the evaluation process and I’m hiring people that meet my benchmark, not some third-party benchmark,” he says.
Pelican3 also is posting invoices to his general ledger – “an AI timesaver” – and is working with a real estate conglomerate that will use AI to keep the buyer informed of the status of the purchase throughout the process and can update the broker on its to-do list and commission schedule to close the deal, Hugenberg says. Another client has created an avatar that “talks” to prospective buyers when they enter a property and provides information about area amenities such as schools and restaurants.
ECMSI also sells security camera systems that use AI to help customers keep track of trucks, people and other items, Galioto said.
“We’re pushing it for our customers,” Blanco says.
The AI-enabled cameras also can flag when someone isn’t wearing safety glasses in an area where they are required, Galioto says.
Risks
“There are security concerns when you put AI out there because AI now has access to all your data. You have to make sure you know how much you’re feeding it,” Galioto says.
If AI is turned on inside the infrastructure and the data is not secure, someone could type in a request for all available payroll information, Blanco adds.
“It’ll scan your whole network and pull documents that have payroll information and puts the company obviously at risk,” he continues. “So, AI has forced companies, if you’re going to implement it internally for data purposes, to really clean up and secure the data.”
From the user side, navigating the number of platforms that are out there, determining which is best for what application and evaluating each platform’s strengths and weaknesses are among the challenges AI presents, Merva says.
“For example, ChatGPT is obviously probably one of the more popular ones. It’s fairly ubiquitous. It’s fairly easy to access and use,” he says. “I don’t personally find it very good at doing certain things, like it struggles with image generation.” Also, Microsoft Copilot “suffers from a lack of deep integration into other Microsoft products,” he adds.
“From my perspective, it’s about, you know, data security and company policy,” he continues. A client or employee could “just go sign up for a million different AI tools if they wanted, and dump company data into that sometimes without thinking of it, and that’s a real challenge.”
Hugenberg points to two big risks that with AI.
One is hallucination, “when AI makes something up,” he says. For example, if a process has a set of known steps and one is left out, it will make up the missing step.
“So, when you’re creating content that is going to go out into the business world or go out to an audience, evaluating the product of the AI for hallucination is really important,” he cautions. “You need to understand how to write prompts, correct the issue and validate.”
Another risk is the integrity of the data begin used. “If I put in bad data, then the user is going to get bad data out. They’re going to make wrong decisions,” Hugenberg says. “So, AI has a data integrity concern that has to be a part of anything that you do.”
“We have not seen the kind of data analytics that I think AI will be beneficial in use for the next one to two years,” Merva says. “That’s where I really think the sweet spot is going to be, and what I kind of hope to see in the next maybe six months or a year.”