Latest News

New Australian robot helping companies make complex decisions

South Australian company Complexica has developed a robot with an algorithm-based persona, which is being used to help businesses make data-driven decisions in real time.

Larry, the Digital Analyst, is made up of a set of algorithms tuned to complex problems in order to quickly generate answers that would otherwise take people a very long time to work out.

In one example, Larry helped formulate a 52-week promotions plan for a national company with 25,000 products sold in 1400 stores across Australia based on the question of which product should be on promotion at which time of year and at which price to maximise profits.

“It (usually) takes about 30-man days to come up with one plan because you are dealing with 25,000 items over 1400 stores for 52 weeks – it’s like a really big Sudoku,” said Complexica’s Managing Director Matthew Michalewicz.

“But in 60 seconds Larry was able to consider about 10 million combinations of different prices, products, frequency, and predict how much more you will sell with all of these combinations and convert them into weekly averages per state and per store.

“A machine and all the computing power that sits in the cloud can consider things that an organisation will never have time to consider,” he added.

According to Michalewicz, Larry is best suited to large companies that experience repetitive sales of everyday items such as food, hardware and liquor.

“Businesses that have complexity are going to get much greater benefits from Larry than businesses that don’t. We define complexity by three core things: how big a business is; how many products you sell; and how many customers you have.”

Complexica has signed up 20 companies across a range of industries and aims to scale up to 100 clients within two years.

Current clients include PFD Food Services, Liquor Marketing Group, Leader Computers and Coventry Group.

1 Comment

  1. I can see that with large amounts of sales data showing the impact of discounts and promotions on categories this is possible. The spreadsheet / database approach would require a business savvy experienced programmer days to get a result.

    The old adage “Garbage In => Garbage Out” comes to mind so there would presumably be a lot of effort involved in making sure Larry understands the data as well as the question. I am keen to see where this goes.

Send this to a friend