A venture capital-backed “AI performance monitoring system for factory workers” is proposing what appears to be dehumanizing surveillance of factories, where machine vision tracks workers’ hand movements and output so a boss can look at graphs and yell at them about efficiency.
In a launch video demoing the product, founders Vivaan Baid and Kushal Mohta put on a skit showing how Optifye.ai would be used by factory bosses.
The YC deleted video for sweatshop startup Optifye pic.twitter.com/vCJvm2HTce
— Adam Lerman (@AdamLerman5) February 25, 2025
“Ugh, it’s workspace 17. Workspace 17 is the bottleneck. The worst performing workspace here,” one of the bosses says, while watching a video of a man making clothing in a factory. “Hey number 17, what’s going on man? You are in red,” he says. “I have been working all day,” the person playing the worker says. “Working all day?” the line boss replies. “You haven’t hit your hourly output even once today. And you have 11.4% efficiency, this is really bad!”
“It’s just been a rough day,” the “worker” replies. “Rough day?” the boss says, looking at a calendar full of red days. “More like a rough month.”
Optifye.ai, launched by Duke University computer science students Baid
26 Comments
adocomplete
I don't think the issue is necessarily the tool, but rather the execution. I think everybody can agree that improving efficiency is a good thing, and tools to track/improve it are a net benefit, but the way they approached it was abhorrent.
But seriously, who approved posting that (before it was deleted)?
dieselgate
Dang this was depressing and appropriate to read on HN I guess.
In unrelated I saw a license plate with just “YC” today.
I thought the article was gonna be the following joke:
It’s a guys first night in jail. At 9pm one of the other inmates yells “51!” And all the other inmates laugh and laugh. A moment later another inmate yells “29!” And all the other inmates laugh and laugh.
The new inmate asks the cellmate – what’s going on with the numbers?
The other inmate replies, “we’ve all been in here so long we assigned jokes to a number and just say them instead. Why don’t you give it a try?”
So the new inmates yells – “14!”. And it’s silent
The new inmate asked the cellmate, “what happened, what did I do wrong?”
And the cellmate said, “you didn’t tell it right..”
speak_plainly
Imagine if they had read something like Out of the Crisis by W. Edwards Deming.
Instead of creating a digital whip for shallow (or even outright harmful) managers, they could have developed a QA tool grounded in real, deep thinking—one that respects proven principles of manufacturing.
There’s still time to course-correct, and embracing some of the foundational literature on the topic could make the difference.
readthenotes1
Reminds of the hinges scene in Schindler's List
https://youtu.be/I-yLk8SZAJY?feature=shared
uoaei
I think we should have a reckoning as a society regarding the prevalence and predilections of sociopaths. Patric Gagne is a PsyD with diagnosed sociopathy (technically anti-social personality disorder, I think) who writes and speaks about her experience and she makes a number of good points. One is that we often can't conceive of sociopaths as anything but the stylized depictions we get in media where they are often portrayed as violent animal abusers or serial killers. Her next point is that reframing this image by identifying a disorder she proposes should be called "low affect disorder" will make it more apparent how people afflicted by it consider and navigate through the world to help us more readily address the potential harms associated with that kind of apathy. It would help us all to be able to recognize that kind of behavior to the point where we can box and name it, challenging its ability to cause harm undetected.
One hopeful note is that the apathy goes both ways: harm is just a means to an end, and if we can offer alternative paths to those ends that don't incentivize harm then we'd all be better for it.
This kind of company, as well as the activities of YC generally, is evidence that we have normalized a certain kind of harm through the development of our economic and government institutions. Those with low affect and who struggle in the empathy department are more than happy to live with ideologies that ignore suffering as long as it's separated by one or more layers of bureaucracy. This normalization is essentially sanity-washing for the profit motive.
I don't see anything indicating this trend will reverse in the near future, there's a pathology in society that seems to prevent actual accountability these days.
Etheryte
This is exactly why we need regulations that prohibit this. The upcoming EU regulations [0] seem to at least partially cover this, as employment is one of the high risk categories for AI systems. What that means in detail is still to be seen, but at least the groundwork is already there.
[0] https://news.ycombinator.com/item?id=43152937
phillipcarter
Consider nearly any professional sports team and the relative effort that athletes put in. There's very little disagreeing on objectives our outcomes (score/prevent points, win games) and incentives are directly tied to this (win more games, get more money).
And yet when you listen to what some of the highest-performing athletes say, they'll readily admit they don't go 100% effort 100% of the time. In fact, that's often a very bad thing to do because you can burn out quickly, and then the opposing team who paced themselves a little better starts running over you. However, there are spurts of intense activity where you really do go 100%, and then you quickly dial that back again to make sure you have effort reserved in the tank. Ideally you get down to 0 at the end of the game, but it's also readily acknowledged that sometimes this is out of your control, and often in quite significant ways, like the football bouncing weirdly when it popped out of someone's hands.
All of this is to say that there's a deep obsession in the corporate world around efficient teams performing labor, but when you get into organized sports where there's literal teams fighting for an objective, they don't chase "efficiency" that would amount to "time doing useful things on the field". Such a measure would be ridiculous.
pzmarzly
I know some people working in low-cost sweatshops, where human labour is cheaper than automation, everyone is told they can be replaced within days, and the few machines you may find are usually older than the employees (or better yet, contractors) operating them.
Every sweatshop like that has high turnover rates, and micromanaging bosses that… let's say make sure these rates don't fall.
If these bosses are the target audience, then I guess the ad is well made? Identify bad employees faster so you can hire better ones quicker, yada yada. I can imagine how this promise can make someone want to buy the software, so fair play for the ad creators, I guess?
I really hope this project fails though.
minimaxir
The surprising thing here is that the startup/YC deleted the video. If the startup was based in San Francisco they'd be running billboards bragging about the productivity boost.
https://news.ycombinator.com/item?id=42380054
beepbooptheory
You could stick to your guns and defend the company your supposed to be supporting, you could not fund these people to begin with, or you could admit a mistake and be honest that it was one. To not do any of these seems like the absolute worst. Morality aside, you'd think YC would at least enough money to hire a good PR person to respond to a reporter with something at all for moments like these!
You gotta really wonder what they are going to teach you at the "AI Startup School" they keep advertising at the bottom of this site..
imglorp
We were told the machines would free us from repetitive or injurious work, letting us pursue more meaningful and prosperous lives. Or at least they would be our partners we could leverage.
Instead, will they be the tools of our enslavement to The Man?
ChrisArchitect
[dupe] Lots of discussion earlier: https://news.ycombinator.com/item?id=43170850
ryandrake
What I'd love to know is: Assuming the founders pitched this idea for feedback to many people before getting this far, including to friends and family, didn't a single one of those people pull them aside and say: "Wait a minute, maybe stop and think about what you're actually creating here…" Could they find nobody in their circle of advisors who are able to empathize with low paid factory workers or at the very least point out the potential PR downside of this work? What kind of bubble are the founders living in? If I pitched this idea to a random sampling of 10 of my friends, I guarantee all 10 of them would retch in disgust.
some_random
I'm shocked that anyone here is confused as to why this is controversial, to use a popular twitter phrase they were completely "mask off" on what they intended their tech to be used for. The demo was for a garment manufacturing sweatshop in which they identified a slow employee, called him by a number, and humiliated him, stopping just short of recommending a "corrective action".
This easily could have been spun in a positive light, imagine a commercial where they use their technology to discover that an employee was using a broken machine or that there was a bottleneck farther up the assembly line that could improve rates! But no, it was a cold look into how sweatshop operators view their workers.
fullshark
The upcoming AI powered laborer surveillance will not stop at sweatshop employees btw.
neilv
I disapprove of this application (knowing how these things will be used), but I hesitate to condemn whomever worked on it, without knowing more.
A few background thoughts…
I remember browsing angel.co, and most of the startup blurbs were just depressing — either crypto scams, or people who sounded unaware that they were doing something bad for society. Many worse-sounding than this one. So depressing, that even LinkedIn was a more pleasant place to be scrolling job listings.
But there can't be that many sociopaths, as it appeared there were on angel.co; they're probably mostly just people with not much world experience, and perhaps too much techbro echo chamber time, now going through the motions they were taught about how to be a startup founder.
Also, there's modes based on context, and I'll embarrass myself here… I remember during the dotcom gold rush, when I'd already had wholesome values instilled in me, plus the usual Internet-before-it-went-corporate values, yet I found on occasion I got distracted from those, in a mode. I was a grad student with a research assistantship, in a lab that had sponsors and VIPs coming through every day. I recall one time some people from a particular huge music industry company came through. So, I start telling them something about the emerging democratization of creation and distribution of high-quality music… and how I thought their company can get ahead of that. Afterwards, I was, like, why did I go out of my way to do that, like an amoral hired gun? On any other day, I'd be trying to help people to be empowered, and encourage them to get rid of that same company.
Going back to this particular application domain, another thought… If you've worked with factory people, at least the ones I did (in a startup introducing a new factory station) were delightful, and I can't imagine imposing this system on people there. Maybe that's another reason to know your customers and users: not only understanding the problem domain and getting the MVP right, but also to engage our natural human empathy and caring.
mattgreenrocks
I believe there's a bright future for software that enhances and amplifies our humanity in different ways. What form would that take? If I knew, I'd be working on it, but I sense a lot of opportunity in un-disenchanting technology for various demographics.
As for this tech: it should die in a fire. If the creators read this, they should understand that they are potentially inventing the shackles that they themselves will be bound in (albeit later).
gowld
It's giving https://www.monticello.org/slavery/online-exhibitions-relate…
> Many slaveowners, including Jefferson, understood that female slaves—and their future children—represented the best means to increase the value of his holdings, what he called “capital.” "I consider a woman who brings a child every two years as more profitable than the best man of the farm," Jefferson remarked in 1820. "What she produces is an addition to the capital, while his labors disappear in mere consumption."
HenryBemis
GDPR doesn't allow any of that garbage.
jagged-chisel
>For the workers? They get the tantalizing benefit of being “held accountable for good or bad performance.”
No. They will not be 'held accountable' nor rewarded for good performance. The reward will be the ever-rising bar on performance until management is satisfied that they will never be required to pay out on those motivational performance bonuses and that workers fear for their jobs.
krunck
"Optifye.ai, launched by Duke University computer science students Baid and Mohta, is backed by Y Combinator, according to the company’s site. On their Y Combinator company profile, they write that both of their families run manufacturing plants, where they’ve been exposed to factory working conditions since they were children. “I've been around assembly lines for as long as I can remember,” Baid wrote.
Mohta wrote, “My family also runs several manufacturing plants in various industries, which has given me unrestricted access to assembly lines since I was 15.” "
So these guys come from families that run factories and manage workers. They NEVER worked in one themselves. It's their turn to sit in the factory for 12 hour a day, 5+ days a week and have AI assisted asses badgering them all day.
Workaccount2
Here is a painful truth that I would venture that 95%+ of people don't know. In fact when I learned it, removing this small block of knowledge from my head toppled an entire ideological castle I had lived in until that point.
Low wage labor is not very valuable at all, and has very low margins on it. The gap between "profitable worker" and "unprofitable worker" is the absolute smallest at the bottom. These workers keep almost all value they create and are never in a position to generate massive value. They are perpetually right on the threshold of being unprofitable to employ.
So the problem then becomes "How do you keep a worker above that profitability threshold when they are barely above it?". You can have kludgey borderline inhumane approaches like this, or maybe try to use perks as coercion to hit targets, but not matter what it's a very difficult problem to solve.
Everyone jumps to "Pay them more" but almost everyone is unaware (and certainly unwilling) that that would necessitate cutting money from their checks to buffer the profitability threshold of these bottom tier workers.
claudiulodro
This is exactly like that famous tweet about the torment nexus[1], but they're trying to make the factory from Charlie Chaplin's Modern Times[2].
[1] https://knowyourmeme.com/memes/torment-nexus
[2] https://criterioncloseup.com/2015/06/14/modern-times-1936-ch…
hector126
> Mohta wrote, “My family also runs several manufacturing plants in various industries, which has given me unrestricted access to assembly lines since I was 15.”
Color me not remotely surprised.
adolph
In case someone hasn't read it, here is a link to Marshal Brain's "Mana" which walks the reader through the creation of a dystopia through worker monitoring/feedback systems called Mana. In Brain's US-centric vision, this system is first deployed in a fast-food restaurant.
Edit: to add a discordance btw Mana and Optifye.ai's vision, for 404 and Optifye, the tool is used by managers to browbeat workers; for Mana, the managers were among the first to go. Why bother with a human to say “You haven’t hit your hourly output . . . this is really bad!”?
https://marshallbrain.com/manna1
toomanyrichies
Late-stage capitalism is watching the movie "THX-1138" and getting inspired to write a YC pitch.