Tech Talk - AI and Machine Learning

By Marty Thomasson | 8/17/2023

Tech Talk - AI and Machine Learning

The advent of AI and machine learning has taken the world by storm. Where there’s a lot of confusion around what it can and should do, there’s also a lot of excitement around the opportunity it brings to businesses. Marty Thomasson, CEO and Founder of Gearbox, sat down with Co-Founder and Chief Technology Officer, David Nahodyl, asking him his thoughts about these emerging technologies and the potential effects on their clients.   

Marty: What are some of the emerging technologies that you’re excited to bring to our clients? 

David: I think artificial intelligence (AI) and machine learning are technologies that are really starting to come into their prime. The public is just starting to recognize the power behind these tools. As the technology advances, I think businesses that take advantage of them will find themselves with a competitive advantage in their industries. 

Having custom software for your business is powerful, but when you bring in machine learning, it becomes a force multiplier. So, I think it’s going to be a huge advantage for us to start implementing AI and machine learning into solutions for our clients. 

Marty: Can you clarify the difference between AI and machine learning?

David: AI or artificial intelligence is software that mimics the way humans think to perform complex tasks. Machine learning is a subcategory of AI that uses algorithms to learn new insights and patterns from data and then apply that learning to make better decisions. They are used interchangeably a lot, but machine learning is just one aspect of true AI. 

Marty: What are examples of projects we have implemented using AI? And what business problems have we solved using these technologies? 

David: Well, one great example is a client of ours who takes photos of signs for office buildings, hospitals, retail spaces, and more. Historically, when the photos were taken, it used to require a lot of time to manually sort them to identify which ones would have to be replaced. By implementing a custom-trained image recognition machine learning model, we were able to provide a solution that would categorize the images in seconds. This frees up the inspector to move on to their next job instead of performing a time-consuming task. 

Marty: What are some other use case scenarios, particularly within the general contracting and specialty contracting industries, where you see big opportunities? 

David: Many contractors have people in the field who need to write a report that will be seen internally and by clients, like updates, and such. They need these field reports, but they also should be clear and concise, and maybe English isn’t their first language. They can use assistive technologies like AI and Chap GPT to generate and help clarify these reports. 

Marty: So, it can be a great assistive tool for their jobs?

David: Yes! I think a big fear that many employees share is they are worried about technology replacing their jobs when in fact, it can enhance their jobs in ways that will make it easier and more productive. People once feared computers, and they wound up creating many more jobs than they ever replaced because they freed us up from mundane tasks and allowed us to explore new opportunities. 

Marty: What’s another advantage from a corporate perspective? 

David: Here’s a good one. Machine learning trained on your own company’s data can help you recognize cases where you might be underbidding on something. Or let’s say you have a 300-page document to read. It can do things like help you recognize areas in floor plans or blueprints where you might have missed an area on a diagram that is going to require attention or might require extra effort. It can even assist with labeling and content validation to make sure forms have the right kind of information in them. The use cases just keep expanding. One can see how the technology can pay for itself while helping a company deliver a higher-quality product. 

Marty: Makes sense. Can you talk a bit about the requirements to implement, to take advantage of these technologies in custom software?

David: There are a lot of pre-trained models that we can use that don’t require additional data from the client. Just plug and play with minor tweaking. Most would benefit from these without having to make a huge capital outlay. For more specific types of client data, it might take a little more time and investment to train and develop, but there’s a big range. 

Categories: Tech Talk