AiFi – Making Retail Store Autonomous using AI
AiFi uses AI, machine learning, computer vision and sensor flow to bring a check-out free experience to any retail store concept.
AiFi is an AI and computer vision company that uses AI, Machine Learning, Computer Vision and Sensor Flow to bring a check-out free experience to any retail store concept. Two years ago Amazon launched its own AI/ML enabled store Amazon Go1, where you just walk into an Amazon Go store, pick the items you want to buy and walk out and a bill is automatically sent to your Amazon account. AiFi is aiming to bring a similar experience to any retail store concept from small convenience stores to large department store such as Walmart.
Amazon Go shopping experience
Today retail sales are still overwhelming through retail stores and according to Steve Gu, AiFi’s CEO, last year Americans spent over 37 billion hours waiting in line which is one the main frustration retail shoppers face.2 The AiFi solution intends to make the shopping experience more efficient and convenient. The CEO further states that “AiFi is designed to scale – it can support tracking up to 500 people, and tens of thousands of SKU item numbers.”3
The AiFi solution does not only allow for a checkout free experience it will allows shop owners more collect data on individual shopping behavior by monitoring and analyzing movement, gestures and actions in the store. Ultimately, this will allow retailers to analyze shopping behavior like never before. What further makes the proposition unique is that it does not require custom layouts or major retro-fitting, but could be installed in current stores.
The following are the main advantages I see of the AiFi proposition:
- Cost saving for stores
- Improve customer experience translates into higher sales
- Generate Insights to retrofit stores
- Get real-time alerts on stock outs
- Understand consumer behavior and shopping patterns in store
- Stores could remain open 24/7
This proposition similarly has a few disadvantages:
- Errors, potential errors in image recognitions, processing or an unusual scenario for which the AI is not trained
- Consumer privacy issues
- Potential regulations on privacy and Retail Autonomous stores
- Job-losses that could result from the displacement of Cashiers and other store employees
- Expensive to implement, high-upfront cost, projected to reduce as key hardware products become commoditized
AiFi has three products right now:
- The Autonomous Store Platform:
This is a Turnkey solution which can be used in existing stores to enable them to become autonomous, integrating several of AiFi’s APIs and seamless integration with existing POS, front and back-end systems of the retail store. This is enabled by computer vision algorithms that create a 3d map of the store and tracks every movement from entry to exit of a customer.
- NanoStore
NanoStore is a custom designed and built fully automated, container sized, checkout free store that could be operated 24/7 without store employees.
- The SimStore
SimStore is simulator for autonomous stores. It simulates retail environment to train the AI for different scenarios, which enables deployments to handle the whole host of real world scenarios.
AiFi leverages Synthetic Data to Train AI
The main challenge that seed and Series A companies like AiFi face is to train AI and develop machine learning algorithm requires sourcing large scale of data which is extremely costly. AiFi is navigating this challenge by using synthetically generated data, simulating stores using Avatars to generate synthetic shopper behavior data to refine and train its AI4. Gartner projects that by 2022, 25% of AI’s training data will be synthetically generated4. The approach of using synthetic data is used by Waymo for the driverless car project while Tesla has largely developed a real world data set4. The advantages of using Synthetic data include: Cost advantages, generate much higher volume of data, and protect consumer privacy.
Overall, the AiFi proposition seems extremely promising which has the potential to completely revolutionize retail using Artificial intelligence, Machine learning and Computer Vision. With more data collection the algorithms would become better and creating large barriers for entry for new entrants.
References
- Amazon.com. 2020. Amazon.Com: : Amazon Go. [online] Available at: <https://www.amazon.com/b?ie=UTF8&node=16008589011> [Accessed 17 April 2020].
- Coberly, C., 2020. Aifi Aims To Bring Checkout-Free Shopping To Any Physical Retail Location. [online] TechSpot. Available at: <https://www.techspot.com/news/73463-aifi-aims-bring-checkout-free-shopping-any-physical.html> [Accessed 17 April 2020].
- Perez, S., 2018. Https://Techcrunch.Com/2018/02/27/Aifi-Emerges-From-Stealth-With-Its-Own-Take-On-Cashier-Free-Retail-Similar-To-Amazon-Go/. [online] Techcrunch.com. Available at: <https://techcrunch.com/2018/02/27/aifi-emerges-from-stealth-with-its-own-take-on-cashier-free-retail-similar-to-amazon-go/> [Accessed 17 April 2020].
- Gartner, 2018. Predicts 2019: The Democratization Of AI. Gartner, pp.8-10.
This is so cool! Thanks for sharing this company.
My question is around value capture. How effectively is AiFi capturing the cost savings and improved throughput that it creates for retailers? How is it pricing its service? I wouldn’t be surprised if they lost money on sales of the NanoStore offering and the SimStore, but it might be worth it if these technologies give them more user data to train the computer vision algorithms for a the turnkey platform. Platform as a Service (PaaS) businesses are definitely doing well from a value capture and valuation/funding perspective.
Overall, I love the idea of a completely AI-powered store without employees (though it does seem a little creepy!). Will be interesting to see how the model evolves.
Thanks for sharing these insights into AiFi! I especially appreciated your breakdown of the advantages and disadvantages of AiFi and am curious to see how the company will address concerns around privacy and check-out errors. Who will own the data — AiFi, the customer, the retailer, the POS provider? I wonder if AiFi needs to target specific types of retailers that sell specific types of products, so as to limit the range of store layouts at the outset. Especially in a world that is increasingly moving toward experiential retail, I wonder how AiFi can integrate more complex components of in-store experiences for retailers.