Image: ThirdEye team

Unless you have been hiding in a hole, you will have heard about Artificial Intelligence. Thanks to the perfect storm of data proliferation (datasets are required to train an intelligent system); open source frameworks (models can be built on top of these allowing for easy iterative experimentation) and specialist hardware (accelerating training time) AI is coming of age. You may have also heard the scepticism that much of this is ‘hype’ and in fact,only two-fifths of all AI startups in the UK actually show evidence of AI. Luckily for us in our latest investment in Third Eye, we have found a startup that truly takes advantage of the evolution of technology in this space.

When investing in a business that has machine learning at the heart of its technology we are cognisant as investors that there are some key challenges for any startup team to overcome. Firstly, data is king and obtaining high-quality, real proprietary datasets to build models on, is vital. Secondly, lots of the technology in the stack is becoming increasingly commoditised, so having a deep understanding of end-customers’ needs is essential.

We were fortunate enough to meet the Third Eye  founders a number of years ago and since that time, Raz and Peter have built a fantastic team. Third Eye offers an intelligent software solution for retailers which leverages existing security cameras to understand potential instances of in-aisle and checkout theft. In-aisle, they have used pose estimation technology overlaying CCTV video (and a store floor plan) to understand what products customers are browsing, engaging with and picking up. Then at the checkout, they have combined point-of-sale data (time stamped) and object recognition technology to understand what is being scanned…or not. Where there is activity which may need to be acted upon, Third Eye neatly delivers real-time footage to its staff in-store via smartphones or smartwatches. Staff then simply swipe left or right, Tinder-style to confirm or deny whether the alert is ‘real’, helping to train those models to deliver better insights to in-store colleagues.

The team have attracted some world-class engineers but, importantly, it was clear that their technical team weren’t just building an interesting product for the sake of it; they were applying their solution to a very painful problem. Brick and mortar grocery retailers, in particular, are under increasing pressure with more market participants and the rise of online retail. In addition to this, ‘shrinkage’ – which refers to loss of product inventory – accounts for around 1.5% of sales. For context, this would represent c.£500m per annum of one the UK’s largest grocery retailer’s sales. In discussions with customers we learned that the primary reason for shrink is theft, and this problem is getting more acute over time. Notably, the team had worked extremely closely with certain large supermarket customers who had given them very useful insights into their product set. They had validated Third Eye’s assumption that, whilst the store of the future is certainly on the horizon, there is a lot of value to be captured in the short-term. By offering advanced solutions that work within the current retailers’ parameters, using systems integrators and fitting in with server space on-prem, they could empower, rather than replace, employees on the shop floor.

In the long term we are very excited that the product’s potential goes beyond loss prevention for retailers. We can also see that right now, the product responds to a pressing need as the team successfully secures roll-outs in a number of UK and French supermarkets.

We are very excited to lead a £2m funding round into Third Eye . We’re joined by True, a retail specialist investor, by existing investors Episode 1, who led their previous round, Entrepreneur First, Seedcamp and a fantastic set of angel investors, some of whom we’ve been lucky enough to fund in the past.