Artificial Intelligence. Words that are becoming increasingly common to hear, yet 10 years ago, was a virtually unknown concept. It is something I frequently read about in the news and Facebook articles in conjunction with people like Elon Musk and Bill Gates, yet I'm guilty of having no real understanding of it. So what is this vague and mysterious "Artificial Intelligence"?In layman's terms, Artificial Intelligence (Ai) is essentially about programming machines to reason, learn and problem solve as a human would, and in the case of Auckland based company IMAGR, Ai can be about programming shopping trolleys and baskets to instantly recognise food and other items and automatically charge them to your linked payment method as you leave the store. No lines, no check-outs, no hassle.
I had an amazing opportunity to chat to IMAGR's founder, William Chomley, as he gears up to take SMARTCART international.
Natalie: Tell us about IMAGR and the current projects you have in the pipeline
William: IMAGR is an Auckland based Artificial intelligence company that specialises in image recognition technology. Our team comprises some of the brightest hardware and software minds in the country (and the world), which we need for the solution we are building. Our current project is SMARTCART, which is targeted towards retail, essentially it brings the point of sale to the shopping trolley. The purpose of this is to create a truly personalised in store customer experience, by bringing the point of sale to the shopping trolley or basket.
Natalie: Where did the concept for SMARTCART come from?
William: Early 2015, I used to work at a finance firm in Sydney and would buy my ingredients for lunch from the supermarket downstairs. Regardless of time of day, the supermarket would always be packed and I would spend my hour lunch break in line waiting to get out. I decide that there had to be an easier way and started researching a way to allow customers to skip the checkout.
Natalie: Besides reducing queuing times – how will this technology benefit Retailers and Consumers?
William: Although reducing que times is a nicety, the main goals of IMAGR’s SMARTCART is to provide a more personalised in-store customer experience. We achieve this through identifying customers as soon as they enter a store through their mobile phone. From there we have a number of offerings on the mobile device and as there is no more checkout,we are able to redeploy cashiers throughout the store to interact with customers. As a customer, as soon as you walk in you can load your shopping list and depending on which store you are in, will be guided around the store in the quickest fashion (so you’ll never get lost). We have implemented a recipe function so as soon as you have a number of items in your shopping basket the application will start to recommend products to add to make certain recipes. This can also be pre-set to assist with dietary requirements, diets and even budgets. When you leave the supermarket, you will have an e-receipt stored in the application, this can be sent to yourself via email or even plugged into Xero (no more pictures of your receipts).
As a retailer we present them with an improved ability to project, produce, promote and provide.
When we talk about Project, we refer to their new ability to anticipate demand trends, while optimizing and automating supplier negotiating. This has the below effect to their stores and bottom lines;
- 1-2% EBIT improvements using machine learning to anticipate fruit and vegetable sales.
- 20% Stock reduction using deep learning to predict e-commerce purchases.
- 2 million fewer returns per year.
With reference to Produce – we are talking about Automated warehouses and store operations; optimising merchandising, product assortment and micro space which has the effect of a;
- 30% reduction of stocking time using autonomous vehicles in warehouses
We then discuss Promote – Optimise pricing, personalise promotions and tailor displays in real time which leads to;
- 50% improvement of assortment efficiency
- 4-6% sales increase using geospatial modelling to improve micro market attractiveness
- 30% online sales increase by using dynamic pricing and personalisation
Finally, Provide – Personalise tips, suggestions, offer immediate assistance with re-deployed staff and automate the instore checkout. All of which we have discussed above.
Natalie: How long do you think it will take for artificial intelligence technology such as SMARTCART to be implemented at supermarkets on a global scale and what challenges do you think it will face along the way?
William: SMARTCART has been designed in such a way that it has the ability to scale instantaneously. The reason is that we utilise existing infrastructure, for example, we clip our device onto their current trolleys, we use their wifi and just update their backend server. The product offerings in supermarkets also tend to be fairly homogenous, which means once you’ve done one you can do them all. We also don’t require any downtime in the supermarket to get this working. We can install one night and be operational in the morning. In my opinion, this will be a mainstream technology in 2-5 years and it will need to be. Retailers that don’t adapt and upgrade their offerings, risk being eaten by the more progressive supermarkets and even online retailers like Amazon. A good example is Alibaba, who has just purchased one of the largest bricks and mortar retailers in China and plans to remove the checkouts from these supermarkets.
Natalie: How have you managed to grow IMAGR’s exposure and presence in the market?
To start, my main goal was to get in front of as many domestic and international retail executives as I possible could. This involved cold calling, cold emailing, even driving from Dallas to Bentonville Arkansas for a meeting with an executive at Wal-Mart. Now, every month I send out a newsletter to these retail executives, venture capitalists and engineers to keep everyone engaged and informed as to where we are in the development and launch cycle. I am just starting to write my own content for LinkedIn and Facebook and off the back off this am fortunate enough to be invited to speak at events on AI and retail.
However, I think the biggest credit to our growth in exposure and presence is in the team that we are fortunate enough to have assembled. Their talent, hard-work, experience and credentials speak for themselves and as a result people take note.
Natalie: what are the biggest challenges in the industry that you’ve had to overcome?
William: There are four:
- Finding the talent: This has been one of the biggest challenges to date. Everyone is in the market to find Ai, deep learning, computer vision engineers and just engineers in general. We have had to bring people in from overseas, the south island and other companies to work for us. We are also always on the look out for recruits
- Finding a customer: We had buyers for the technology before we started building it. I had tested our market around the world and knew that once I built it we would have a paying customer.
- Commercial grade recognition accuracy: When I first started the highest recognition in computer vision was around 60%, so if you put that into the number of products that would walk out of a supermarket for free, you’re talking about losses of $10bn plus. We have managed to build a proprietary code base that is able to handle north of 98% accuracy in product recognition.
- Real time image recognition over wifi: We require processing to be done over wifi, which given the speed at which some videos download at, it was a tall order. We have found a way to do this through inferencing and packing up our data in a very unique way.
Natalie: There has been a lot of controversy around the responsible development of Ai recently in terms of loss of jobs, the ethics and moral reasoning of machines and the potential threat it poses to humanity itself. What are your opinions on these issues?
William: We are developing a product and service that will only reach its full potential if humans are involved. Our goal is to place the people behind the checkouts on the floor to interact with customers and improve the overall instore experience. In my opinion, Ai is a tool that when leveraged properly will work alongside employees to make them better educated and therefore better equipped in their decision making. Overall it makes them inherently more valuable to a company because of the systems, services and products that Ai is providing. A great example is looking at JP morgan who just introduced an Ai software that eliminates an estimated 360,000 hours of work each year previously done by lawyers and financial loan officers. Contract Intelligence (COIN) uses machine learning to review and interpret commercial loan agreements. The document is able to scan over documents in seconds and is less error prone than human beings. COIN is just the tip of the iceberg for JPMorgan Chase, which is determined to automate some more mundane tasks to help free up their employees and make them much more informed when it comes to decision making.
Natalie: What is next for IMAGR without giving away your secrets? How will you continue to grow and expand?
William: We have just closed a capital round, where we were fortunate to bring on Sage Technologies. Sage technologies is the venture arm of investor and hedge fund manager Harald McPike. We plan to grow exponentially over the next 6-12 months, in our staffing and technology. Our beta will be ready to launch in April 2018, which we are in the process of building as we speak. Following on from there we aim to expand into New Zealand, however we have been asked to be implemented in Asia mid 2018 and be operational in the USA by this time in 2018. There is certainly a lot of international interest at this stage.
Natalie: If you had to give one piece of advice to an up-and-coming young entrepreneur with a dream, what would it be?
William: The business I left my full-time job in finance to pursue failed, and it failed badly, I lost time, my job and my money. It was all because of one fundamental flaw, I thought if I built it, people would come and use it. I was essentially building a solution that nobody would buy or use. Therefore, if I have one piece of advice it would be to shadow test. Shadow testing is the process of selling your idea, service, product etc before it even exists. Approaching your soon to be customer base (and not your friends or family) and ask them if they would pay for what you have. This validation shows you have a product, customer and market to go out and build for and at the same time it also helps manage a bit of the unknown risk that saw me fail my first time around.
Natalie: The future of technology is advancing in leaps and bounds everyday and IMAGR's SMARTCART is one of the best examples of New Zealanders blazing the way as pioneers of innovation. By taking away the inconveniences and inefficiencies of everyday tasks, IMAGR is creating endless value globally by opening up opportunities for upskilling of jobs and an improved ability to anticipate demand trends that retailers can cater to. IMAGR will continue to inspire a generation of future entrepreneurs and i'm very eager to see what they will come up with next. The future is now and it is phenomenally exciting.