What's an algorithm factory?
Source: Alexandra Cain
Berend Berendsen has such good foresight it might not be a bad idea to ask his views on next week’s lotto numbers.
The co-founder and Australian managing director of Widget Brain started studying artificial intelligence in 2001 when people looked at you funny if you admitted that.
But the laugh’s on them because his choice of university course turned out to be particularly prescient.
His business is an algorithm factory. I didn’t have any idea what that was, either, before Berendsen explained it to me.
Essentially, he writes and also white labels computer code that software firms and other businesses can use to help solve problems, answer questions and run businesses better.
Berendsen explains what an algorithm is. “It’s an automated way to execute a number of steps, which can be something very simple or very complex. What we notice when we work with organisations, is that they need to automate or optimise their businesses, and for that they can use algorithms.”
But they have two problems to achieve this. They don't have the know-how to create those algorithms or deploy them. And even if they have those algorithms they don't have a link between someone who can create it and someone who can apply it to their business.
“So what we have done with the algorithm factory is automate a number of those steps, and we provide organisations a full service for algorithms,” Berendsen says.
For instance, a cafe could use an algorithm to look at its data about when it’s busy, to forecast when to roster staff. Or a retailer could use an algorithm to forecast sales or foot traffic.
In the cafe's case, the algorithm and its output may not need to be complex – it may simply indicate two staff are required for the 9am early shift and two people at 2pm. Although a lot of number-crunching may be needed to get to that point, and the business employs mathematicians, physicists and computer scientists to execute its work.
Davey Water Products is an example of a customer with complex needs. It manufactures water pumps for high-rise buildings, farms and golf courses. Widget Brain analyses data the pumps produce to measure how efficient they are and how well they are running, to detect leaks. This helps save water and control costs, which is important because a golf course may not be aware its underground pumps are leaking without this information.
The business has been bootstrapped and built customer by customer and has bases in Canberra, Rotterdam and Burlington in the US. It services about 30 customers and is aiming to have 30 full-time staff by the end of the year. Last year Widget Brain generated $1.6 million, a figure it’s aiming to double this financial year.
Commenting on the firm’s future direction, Berendsen says: “One of our goals is to provide algorithms-as-a-service. We don’t want to employ droves of people, we want to develop a repeatable service.” For instance, it can use the algorithms it has developed to forecast retail sales for a client, then apply the same thinking for the benefit of other retailers.
Berendsen's tips for other entrepreneurs is to be persistent even when people tell you something won't work or you need to change your approach. He also says be wary about accepting investment funds too early.
“Believe in your model and try to make it work without investment first, because that will prove the business model,” he says. “Not that investments are necessarily bad; they are a means to an end. Now we have a stable and growing business, we might decide to do that. But if I were to start a business again, I wouldn't start with investments from day one.”
It’s good advice for start-ups and bigger businesses alike.