Home PublicationsData Innovators 5 Q’s for Per Nyberg, CCO of Stradigi AI 

5 Q’s for Per Nyberg, CCO of Stradigi AI 

by Eline Chivot

The Center for Data Innovation spoke with Per Nyberg, chief commercial officer of Stradigi AI, a startup based in Montreal, Canada, that provides AI solutions to enterprise businesses. In this interview, Nyberg discusses how Stradigi AI’s platform “Kepler” uses machine learning to automate data science tasks, allowing users with no previous experience to use advanced machine learning for key business use cases.

Eline Chivot: What was the main inspiration behind Stradigi AI, and how has its purpose evolved?

Per Nyberg: The founders of Stradigi AI identified the wave of the future, and pivoted the company towards AI in 2014 once they realized the best way to solve many important customer problems was with machine learning. From there, the company started delivering many custom projects. At the time, most people didn’t know what AI was, but as customers became more aware of its potential, interest evolved. During that time, the team realized that the best way to serve clients efficiently was by developing a platform with self-serve capabilities, because continuously building projects from scratch is inefficient, unscalable and unmaintainable. That’s when Kepler started to come to fruition in a tangible and meaningful way.

Chivot: How does the Kepler platform work? If entrepreneurs approach you with minimal AI experience or knowledge, how can you make sure to provide them with a tool they can use?

Nyberg: I’m glad you asked that, because allowing people to benefit from AI who have no previous experience is at the heart of what makes Kepler different. In general, when you want to build a machine learning model, the expert or the data scientist would look at the data, clean the data, conduct feature engineering, decide which types of specific algorithms should be used, do hyper parameter optimization—which is turning the knobs of the training—and train the model. These represent a multitude of labor intensive steps that require in-depth machine learning expertise. Kepler automates all of that, which means you no longer need the machine learning expertise to leverage advanced machine learning 

When we work with a client, all they need is data, an internal subject matter expert, and a problem to solve. Kepler automatically selects the best algorithm for your data, produces a machine learning model, and then you have a trained and deployable machine learning model for production. Depending on the data set, this can all occur within minutes.

Chivot: The Kepler platform has already helped industries and businesses ranging from retail to healthcare. Are there industries which aren’t even thinking about AI right now, but could greatly benefit from it?

Nyberg: We are a very fortunate company in the sense that we have a lot of AI expertise in-house. Our people are global leaders. As we look out into the market, most companies are not as fortunate. You can read every analyst and industry report and they all say similar things about the state of AI. Everyone is interested in it, everyone has done some testing, and yet very few people are using it in production. Despite the hype, most people are not taking advantage of the full potential of AI. There are a number of reasons for this, ranging from company culture to skills. The skills gap is the number one problem: If you have a data scientist, you can start thinking about developing your own AI. If you don’t have a data scientist, it’s almost impossible to start. We took that deep knowledge of the market and really focused on augmenting Kepler’s capabilities to ensure we could serve clients in any industry to alleviate that gap. If your team includes someone whose skills are about supply chain or inventory forecasting, and they know their data but not necessarily AI, they can now reap the benefits of machine learning. As an organization, getting machine learning in the hands of anyone—at any company—is really our focus today.

To your question about industries, there are three dimensions. First, it’s important to note that almost every industry is looking at AI, but some industries are further along than others. Retail might be one, e-commerce might be another, but there is a lot of variation from company to company. So two companies may look identical, but one is using AI, and one is not. There is more variability along those dimensions. It is really about whether a company is prepared to use AI. Second, and generally speaking, industries that are adopting AI, especially in production, are those that can monetize their data. We all give our data away every day on our phones and on websites, so they can use AI easily to monetize that. But industries like healthcare or financial services that have a lot of regulatory burden have a harder time monetizing the data. I think what you are going to see is that more and more use cases will come up and evolve as the regulatory concerns are understood and addressed—we write about that in our recent AI e-book. The third dimension is that—and it never ceases to amaze us—sometimes we meet customers, say in a sector like agtech—agricultural technology—and they are using AI in very novel ways. These are industries which you would think would be very conservative, but are actually advanced adopters of AI.

Chivot: How will AI be used in industry and businesses in five years? What are the biggest barriers preventing more entrepreneurs and businesses from adopting AI? 

Nyberg: I believe the reports when you look at AI being ubiquitous in an organization. AI is not the solution to every problem, but there are many problems where you could apply machine learning. For current problems such as inventory forecasting, something even the very first businesses had to do—machine learning is applied to build better predictions, making those businesses more efficient, and smarter. At the other end of the spectrum, we can also see a set of new use cases that are generating revenue. On another note, there’s also the case of recommender systems, which are one of the most widely adapted forms of machine learning for retailers. But one of the problems of recommendation systems is that they base these recommendations on a lot of data, particularly historical data. But if you speak with various stores, especially ones that are fashion-oriented, they change their merchandise every quarter. That’s how fast some have to run their business, and in some cases they have very few articles. This is an area where machine learning can be applied in very novel ways with very little data and create entirely new abilities for these companies to build recommendation systems. And then, there are all those use cases which simply don’t exist yet.

Specifically, I’m talking about challenges that will be multi-data oriented, using data from different parts of the organization to get this 360 degree view of the company’s health and their customers. The ability to do this in the future will be revolutionary for many organizations.

In terms of the adoption of AI by businesses, the first element is not an obstacle but rather a prerequisite: It is data. The first step for any company to use AI is to become digitally transformed. Once that is achieved, there are often other hurdles—the first one being culture. There is resistance, people may need education surrounding AI, or they are worried about their jobs. That’s often the first thing companies need to manage. This is like any new transformative technology where a degree of change management is required. That may be a non-technical hurdle, but it is a very critical one.

Finally, the skills gap is an issue. Most companies are never going to have enough machine learning expertise talent, so they can’t build everything on their own. They need tools without being able to have the expertise. I think that’s the most critical piece: Being able to scale machine learning in an organization by having more people use it to solve critical, business-driving use cases that help inform decisions and generate insights This is the biggest promise of AI: That it will make every individual in a company better at their jobs in some form.

Chivot: What are your future plans and goals for Stradigi AI?

Nyberg: We are in a very exciting time because what we call the “self-serve version” of Kepler, which is the ability for business analysts to use the platform, is ready for our clients and partners. This is a major milestone. Like most organizations, we were impacted by COVID-19, so our market launch plans changed significantly. Nevertheless, we are excited to get it in the hands of organizations who are looking to solve critical problems and boost efficiencies by making smarter decisions.

You may also like

Show Buttons
Hide Buttons