A quick search online shows that Artificial Intelligence (AI) originated in the 1950s, driven by developers dreaming of a future filled with machines able to mimic and emulate human thoughts, actions, and behavior.   

Fast forward to today and that lofty pipedream is very much a reality. Like it or not, these past few years have seen such a rapid advancement of this powerful technology that AI is no longer simply a buzzword, but the next logical step in all things digital.  

It’s fair to say that AI now permeates almost every facet of our daily routines despite the fact that you may not even know you’re using it. Take, for example, Voice to Text for authoring a text, or completing your sentences while typing an email—or even a “copilot” writing the whole thing. You now see that internet search engines return AI-generated results, instead of just keyword search results. And seemingly all customer service interactions start with a near human-sounding chatbot who can answer more and more questions until a live agent is needed.  

AI has silently been influencing our digital lives for more than a decade—from pricing algorithms for airline tickets to what content is fed to us on social media. And with vast amounts of AI-generated text and images, it is now virtually indistinguishable from “organic” intelligence online.  

Most of the buzz right now is about how Generative AI is automating user workflows, with the large language models like ChatGPT or DeepSeek dominating the headlines. But there’s far more to AI. For example, predictive algorithms have been available for decades and are increasingly being used in planning and optimization use cases where vast amounts of data, variables, and math are needed to more accurately predict what will happen next quarter or next year.  

Is it a change for the better? Many would argue it is, particularly in the business world, where AI is making light work of the more mundane, labor-intensive processes and saving users a lot of time, effort, and money. But today’s focus on AI is to make better business decisions faster. Utilizing AI properly can significantly reduce the likelihood of mistakes due to human error and can easily handle complex processes that have become too data-centric for today’s business decision-makers.  

In the past, for example, manufacturers would depend on regularly scheduled maintenance routines to keep their machinery running smoothly—relying on a schedule recommended by the machines’ manufacturers. However, unplanned downtime is so disruptive and expensive, that AI is being used to predict machine slowdowns or failures based on performance data collected from the machine’s bearings, motors, and electricity draw and applying known failure mode patterns to calculate a predicted future failure. This gives a manufacturer the chance to address the symptoms before a problem ever arises, rescheduling a production run to perform maintenance before a shutdown occurs.  

Other forms of predictive AI are being used in demand and supply forecasting, using data from historical sales, seasonality, past performance and external market signals to better predict the future in order to make smarter investment decisions to support growth strategies. 

Is AI a threat or an opportunity for business?

“Artificial Intelligence should definitely be seen as a fantastic opportunity,” says Lou Zhang, Data Scientist at Datacor. “It has great potential to assist current efforts to transform the industry by automating rote and repetitive tasks as well as accelerating digital transformation.”   

“Data is the new oil.” says Zhang, “but today, many manufacturers have more than they can handle and develop issues with data accuracy and repetitive data entry—often the result of manual data collection and re-entry processes. AI can help with, and improve, data quality through automation of data entry and workflow processing in context of the job at hand.”   

“Optimization is also often talked about in the formulation world,” adds Jeff O’Brien, VP of Product for Datacor. “We can use AI algorithms to optimize a particular production run, or series of production runs based on cost or availability of ingredients. Manufacturers are always trying to ensure high quality at least cost, but the math and number of variables has always limited what traditional formulators—or even software—can do.”  

What are some of the pitfalls of relying on this technology? 

“This is where ‘human in the loop’ is going to be critical for training models and validating outcomes.” says O’Brien. “Maybe you use an AI agent for invoicing and it's only 80% confident, but by the user jumping in and taking over, AI can train itself on how to respond in future scenarios, increasingly raising its accuracy to maybe 95% and learning from the process.” This is a dimension of AI called machine learning and continues to get more accurate over time.  

In manufacturing, accuracy always matters, but depending on the use case, sometimes accuracy is critical or non-negotiable. In accounting, for example, an AI agent may be able to handle supplier invoices that have come in that day, scrape all the text, and match everything with the associated PO. AI may make a few mistakes, perhaps processing those invoices with 90% confidence. For a $10 invoice, you might just let it go as it would probably cost more to correct. However, if it’s a $25,000 invoice, a human will want to check it. That’s where ‘human in the loop’ is critical. When the AI agent has processed everything as well as it can, a human can then take over. By watching how the user responds to that exception, this can be fed back into the machine learning model.  

“The thing about these AI agents is they don’t need to be perfect; they just need to be better than the best available human,” says O’Brien. “Companies that head down this path of using agents to automate user workflows will want to take a crawl, walk, run, approach so they can develop confidence in the agent just like they would a new employee.”  

Can AI accurately do tasks typically done by humans?

“Certainly, especially repetitive tasks like entering data,” says Zhang. “For example, many liquid products are delivered in cylinders that cost more than the product inside them. The manufacturer who owns them wants these inventory items back, but today they rely on manual processes and spreadsheets to keep track of them and must encourage customers and distributors to return them. By tracking these containers alongside the required traceability for the product and understanding customer consumption patterns, AI can help predict how many will be returned and when, eliminating the need to order more to keep up with customer orders.”   

Manufacturers today struggle with skills gaps and labor shortages. Going forward, they can leverage AI agents do the highly repetitive, predictable tasks and free up employees to do what humans are better at—managing the exceptions and doing more innovative work.  

Is AI expensive to implement?

“If your competition is implementing AI to make better decisions, automate tasks, and reduce their costs, then it will cost you more if you don’t implement it,” says O’Brien. “There can be some risk involved, but I would say any investment in technology, whether it's AI driven or not, should always be viewed from the perspective of what is my return on investment? An AI agent that can process all your supplier invoices—and never call in sick—can help you avoid hiring additional AP clerks whom you may not even be able to find in time or quantity to support your growth objectives.”   

Can I trust this technology?

“You shouldn’t simply trust the resulting output of any new technology—especially early on when you're working with AI for the first time and training it on your data,” says O’Brien. “But in context of enterprise software, we're talking about narrowly focused agents that will be embedded deep within well-defined business workflows. For an agent that’s processing a supplier invoice—and nothing more—there’s very little risk. Plus, with ‘human in the loop’ you have a contextual filter.”   

It’s important to not think of AI as a threat. It should be seen as an opportunity for the business as well as all the employees. It can be concerning for people who don't really understand what AI is, but that's really a change management exercise—which is for bringing on a new product or technology. AI is not about trying to replace people; it’s about freeing up people to create continuous improvement opportunities.”

How can I prepare my business for AI?

It’s always important to train your employees properly and let them know that AI agents are there to ‘scale’ their output, not replace them. You should highlight that AI is just the next generation of computing, just one that can identify more patterns in bigger data than a human can, and one that can simply make them more productive in this data-driven world.   

“Everything AI related, whether it's generative, predictive, or optimization based, is only as good as your data,” adds O’Brien. “Good data governance practices need to be in place. You should be periodically going through and reviewing your data, making sure it's still accurate and up to date – and if it's not, putting a plan in place to clean it up.”  

“At this point, before you go and hire someone, you should already be thinking about whether your existing team could effectively do the job with the assistance of AI,” adds O’Brien. “In the near future, I fully expect that entire departments like accounting, procurement, or customer service could be run with a single human overseeing multiple AI agents.” 

TrackAbout is actively incorporating AI into our software to help our customers boost revenues, reduce costs, and improve productivity.