introducing bram wiggers from viaconnect group.
Jasper de Vries, Parth Tiwary, Lennart van der Guchte, Bram Wiggers. The four members of an agile and close-knit Data Science and Artificial Intelligence team operating independently of, but within, ViaConnect Group. ViaConnect Group in turn helps companies optimize performance, improve company culture, develop software and implement technology.
As a team working more akin to a startup, the group enjoys the flexibility, focus, rapid progression and lack of restrictions that come with a small, concentrated setup. They’re also passionate about learning, something that is fundamental to their success, and their output reflects this; working on multiple projects simultaneously, iterating quickly, taking on feedback and moving rapidly towards products that solve real-world problems.
Indeed, the idea that underpins all of the team’s work is using machine learning to add value to people’s lives whilst also making a positive impact on society.
We talked to Bram about why learning is so important, the ups and downs of small teams, the benefits of being connected to ViaConnect Group and the problems with company missions.
Your team is made up of just four people, but you’re also part of the larger ViaConnect Group. Could you talk a little about that structure?
All of us within the team value the Silicon Valley-type startup for its flexibility, its self-directed focus and the lack of bureaucracy. At the same time, however, the experience, network and infrastructure that come with larger companies is also valuable. So, by operating as a small, agile team within the larger ViaConnect Group, we can benefit from both structures.
ViaConnect Group also benefit from this setup, as we can solve problems rapidly and provide quick feedback. Our solutions also have a very ‘warm’ way into other big companies because often our colleagues are already working with those companies. And as a perhaps overlooked benefit of working within a larger company, it’s just nice to have a basic company infrastructure in place when you start; laptops, working places, salary, administration etc. It’s comforting to not be distracted by these things and be able to focus on the things we’re good at.
It’s also exciting to present new products to the sales teams of ViaConnect Group. We put a lot of effort into the technicalities, but when we have feedback sessions with sales, they view the product in a totally different way. I think this diversity of perspectives is crucial when building a product.
I think it’s also important to mention that often there’s a sense of safety being part of a bigger company, but that does risk making you lazy. Luckily for us, however, because we work autonomously and have to deliver products to ViaConnect Group, we don’t feel that safety as much, and we’re constantly having to prove ourselves.
How did the team come together?
That’s a tale [laughter]. Jasper was at another startup working on a project to optimize a data center. ViaConnect Group took over the project and Jasper transferred with it. Then they needed more Data Scientists to work on the project, and that’s where Parth came in. He transferred onto the project in September this year. Meanwhile, Lennart and I were looking for a job. We knew each other from Groningen and discussed how the job hunt was going, and we found out we were applying to the same job at ViaConnect Group. Instead of making it a competition, we called Jasper to ask whether we could be hired together and start this AI team. When we met Parth and found out he too had recently graduated in Groningen, everything fell into place.
What’s your team’s mission?
The problem with missions, targets and impacts is that they can be victim to marketing cliches — ‘Corporate Social Responsibility’, ‘People Planet Profit’, ‘Real Impact’. It’s hard to find a description that doesn’t feel like it’s made in a PR office, whilst never really getting to the core of why a business is doing something.
For us, however, our aim is to “develop scalable end-to-end machine learning applications which add value globally and make a positive impact on people’s lives, organizations and society as a whole”. We’re passionate about this aim and delivering a product that functions in the world. We don’t see value in applications that miss the mark on their purpose whilst fulfilling a marketing aim. That’s nothing against marketing, but we believe marketing has to come hand-in-hand with a real solution, as today it’s mainly a zero-sum game.
So, that’s the idea. We want our solutions to apply to many, and not just to the needs of a single business. Sometimes this works, sometimes it doesn’t. We hope that by developing better products, the choice for customers becomes bigger and our possibility to share our worldview can be reflected in that choice. We actually dedicate our Fridays to projects and products that aren’t steered by a client. This way we can develop our own products which fit our view of ‘value’.
Could you briefly describe the projects you’re working on at the moment?
Currently we’re working on two products of our own and two client products. The first product is called Summi. Summi is a system that listens to conversations or calls and extracts the important information out of it. Even though a business makes hundreds of calls every day, it’s hard to extract more information from it than the employee is entering after the call. To gain valuable insights from these calls, we transform the sound to text, and then Summi finds a summary, important addresses, phone numbers and topics from the call.
The second product is a benchmarking tool for the energy usage of buildings. Building managers can enter numerous features of their buildings and see how they’re performing. Currently, buildings are judged on their ‘Energielabel’, but that’s not based on actual energy consumption. So, we’re fixing that.
For one client project, we’re assisting a data center to optimize their energy usage. They have a lot of knowledge in the field of engineering, and we know a lot about combining different data sources. This combination of engineering and data science makes for a very interesting project.
Lastly, we’re helping a plant growing firm with demand prediction. Knowing demand is essential because of growing times, and if demand is predicted incorrectly, vast amounts of stock can go to waste. We’re helping this company assign how many m2 should be reserved for each plant to fit demand perfectly.
So you operate by experimenting, testing and developing numerous projects simultaneously — a process of continuous learning. Why is this important?
Everyone has their own biases. I might think we can help the world by developing the most sophisticated models, someone from sales might want to find the product that will sell the best and as a team we might want to work on the project with the most societal value.
Eventually, however, it’s about the people who are going to use the product. What do they want? Are they eager to never have to transcribe again? Do they want to have more insights into the energy they’re consuming? By testing various projects with potential users, we discover their eagerness to work with our products. We also uncover the shortcomings of our work and we learn from their feedback. Then we address those shortcomings until we realize the product that people want most.
More people are realizing the potential of Artificial Intelligence, but what are the main hurdles that AI has to overcome to realize that potential?
I think the main challenge is that humans find it difficult to change behavior. And in a business context, that makes it hard for Data Science and AI to realize its potential. We encounter many organizations that have this silo structure, where each department is only looking at their own goals and processes instead of having a view of the company’s goals. But this results in a lack of usable data — when every department, team and person is keeping their own excel file, it’s hard to find the possible points of optimization. So, it’s a behavioral problem and a structural problem, and changing that is hard.
How can people learn more about the positive potential of Data Science and AI? And how can teams like yours help facilitate that learning?
It depends on your eagerness to learn. That has to be there, but I’d encourage everyone to at least be open to these topics, even if they can be a little daunting. You could start by simply following someone on LinkedIn like Maarten Sukel who’s doing fascinating AI projects with the municipality of Amsterdam. Or Pascal Bornet, who’s showing what these technologies can do across the world.
To move away from your timeline, there are hundreds of online, informative talks or you could join a webinar. Many interesting lectures, for instance, are given for free on meetup.com. And, of course, you can always just email me to discuss your ideas and fears of AI.
What’s the best outcome of working as such a small team?
The best outcome is the culture. I love starting work every day and tackling a new set of challenges with the team. We’re friends as well as colleagues, and everyone is comfortable asking for advice as well as giving advice. It makes it a healthy, energetic learning environment where we can progress as individuals and as a team. And because we’re so small, we’re also quick to find the team’s shortcomings. That allows us to look for people that will fill these gaps when the time is right.
And what’s the most challenging?
The learning curve. In a big company, there’s a cloud expert, a modeling specialist and a UX designer. But we have to learn everything. Saying that, we love learning, but it’s challenging.
As a small team working within a larger community of startups and entrepreneurs, what are hoping to learn on your journey at B. Amsterdam?
It’s great to feel the energy at B. Amsterdam. There’s a lot of very talented people here who are trying to have a positive impact on the world. We’re excited about having coffees and drinks with them to find out what’s motivating them and to see if we can work together.