CEO faces fraud charges after allegedly discovered AI shopping app using humans
The Justice Department has accused the founder of deceiving investors because his AI shopping app was found to be nothing more than a few people wearing windbreakers. Nate offers a mobile app that assures users that there is a click checkout experience on any e-commerce website “no manual intervention required,” which actually means they can store credit card and shipping information on each website instead of entering it. But prosecutors said the company relies almost entirely on human workers to complete the checkout. Oops.
According to the indictment, Nate’s actual automation rate is actually zero. Instead, the company relies on hundreds of human contractors in the Philippines-based call center to actually complete the deal. Perhaps through “AI”, the company actually refers to “Asian Intelligence.” Nate raised more than $50 million between 2018 and 2021.
The information first reported the allegations against Nate and its CEO Albert Saniger in 2022. Here is an excerpt from the report:
To automatically handle transactions on retailer websites without the help of humans, Nate’s software needs to figure out how to find specific buttons on the page, such as adding items to one of the shopping carts, without being blocked by trackers looking for automated robots, two people who directly understand the technology. It turns out to be difficult.
As a result, Nate’s transaction share manual processing ranges from 60% to 100% throughout 2021.
When Paul Hudson from Glade Brook ordered a pair of Levi’s jeans on the denim website using Nate’s checkout, Saniger wrote in a company’s Slack channel called “Vip-Notificaiton-for-Albert” reminding colleagues that the deal needs to be processed immediately.
Many companies in the AI field are still relying on humans to some extent, especially the data generated by verification algorithms, hoping that AI will take over completely soon. Popular startups like AI are notorious, relying on farms with cheap labor to label data and correct output from AI models. It is reportedly a startup that says it uses AI to automate the process required by law Fivup reportedly relied on humans late last year to correct AI-generated texts. By far, the most successful AI startups are startups that make tools, and the mainstream adoption of AI by enterprises remains a question mark on the concern about reliability.
Nate seems to have entered the hot water place explicitly claiming that humans have almost never been used during the checkout process. Maybe the company thinks it can “forge until you make it” and just runs out of time to get the product to work as expected. But it also enters an incredibly crowded space with Amazon’s dominant e-commerce and other companies, such as Shopify, introducing its own one-click checkout product. It’s also hard to see how Nate grows its business around relatively simple features. Similar companies, such as Bolt, another key checkout startup, have only managed to generate revenue.
Nate seems to be still there, relying heavily on commissions to try to grow its app, allowing influencers to create lists of their favorite products within the app and get any cuts to sell.
Interestingly, however, Nate may have used the product. Amazon, OpenAI and Google are among the companies that recently launched “agent” promotions because they are able to automate tasks including users’ online shopping. These products are still very new and users complain about them being slow, expensive and off-road vehicles. They do present some potential that can automate difficult tasks and make demographics like older adults easier to access calculations.
Saniger faces one count of securities fraud and one count of wire fraud, both of which can result in up to 20 years in prison.
