"Expensify has admitted that its declared AI product SmartScan, which is assumed to scan the expense receipts and categorize the details into corresponding expense pool through a machine process, was actually assisted by humans. Breaching privacy of users, the receipts were posted on freelancing websites where freelancers used to take out extracts of information from receipts and send it to Expensify team."
[0] https://mantra.ai/blogs/pseudo-ai-when-humans-do-bots-work/
I have to concur - i've had the ability to steer companies away from their product.
Expensify was OK before all the OCR stuff and shady billing.
Now with the S1 filing, whatever good designers and engineering talent they had will bail, and the company will be a hollow carcass of its former self. Avoid.
NOTE: the shady billing practices are traditionally the precursor to an S1 filing. Usually, a company going public has earnout/incentives, so they will do whatever it takes to increase revenue preIPO, even if that directly impairs the company's long term future.
Can you cite this? I couldn't find anything seems obscure to search though.
Seems insane to me that a receipt leads to funds stolen.
At the time, I was also amazed beyond words at what Expensify and similar companies claimed to have achieved with their "AI". I find it somewhat comforting to know that they lied and it was mostly manual!
So often I find these 'AI' systems are just smoke and mirrors. You have a great system, nobody is interested. Sprinkle on some 'ai' magic beans, and people are now throwing money at you.
When I looked into it, it seemed as though they restructured their plans multiple times during the ~4 years I was paying them, and I got placed into the most expensive option which was originally the same price as what I started with. I was also locked into a year long commitment at some point, and they wouldn't budge on letting me downgrade to the lower plan mid year (I didn't have a use for any of the advanced features).
The support chat told me the same thing about the changes being announced via marketing emails, and said they could not credit, cancel or downgrade my plan. The support agent also couldn't care less when I said it would cause me to cancel my plan at renewal.
The whole experience left me with a really bad impression, I went from a huge advocate to telling everyone to avoid them.
Ps. I remember their CEO (David Barrett) back from the p2p-hackers mailing list days. He ain't one of them Zuckerbergs. The exact opposite in fact.
As an example, he really believed it was fine to have mechanical turk freelancers see people's receipts. He would say something to the effect of "people throw their receipts in the garbage where any sanitation worker can see them - this no different." I wasn't around when the mechanical turk stuff made the news, but it seems like it caught him totally off guard that people would be upset.
I suspect that the conversation around the price increases went something like:
David: There, I just sent a 3-page-long plaintext email explaining our price updates to everyone. Other Expensifier: Shouldn't we notify people who have opted out of those emails? David: If they opted out of announcement emails, they don't want announcment emails. Seems straightforward to me. Other Expensifier: Yeah, but, like, we don't have multiple opt-out channels. They can either get every multi-page plaintext missive you send, or they can opt out and miss significant price increases. David: Nope. They made their choice, we should respect it. Multiple opt-out channels are an antipattern. Anyway, the number of users who will be upset about this are vanishingly small - let's worry about the rest.
I mean, yeah, he certainly had some terrible opinions about code - that .NET one included. He was also pretty anit-automated-testing, which drove me nuts. But just because he had some bad opinions doesn't mean he wasn't a serious technologist. Perhaps a good coder, but a bad software engineer?
I am not saying otherwise, just objectively speaking: money changes people.
The thing that pissed me off the most is when they emailed some unsolicited BLM propoganda out to every one of our employees.
When I discovered the higher rates we were paying I reached out to support and they said I wasn't notified because I elected to opt out of marketing emails. They wouldn't issue a credit either.
I threatened to cancel service and they truly couldn't have cared less. So we dumped them. Very satisfying. It's a shame - the product was ok. But from my persepctive, F these guys.