Accidental Architects: An Anonymous Tech Worker on Idealism, Disillusionment, and Why Silicon Valley Can’t See What It’s Building
"By the Time the Power Dynamic Shifts, the Culture is Already Set" — A conversation from inside the startup
At 10 pm on a weekday, a software engineer at a small AI startup in San Francisco sent their boss a text message that could have waited until morning, but didn’t.
They’d been biking home from work, moving through the city’s Victorian streets amid the silent drift of self-driving vehicles, endlessly turning over an idea that followed them to the doorstep. By the time they sat down, they had to send it.
“We need to be doing more community-focused work,” the message said.
Not just evaluating AI in finance and math, but actually testing whether it serves the people who never get a seat at the table where these systems are built: teachers, people applying for public benefits, and communities that will live with the consequences of decisions made without them.
The boss wrote back. He was validating and encouraging. Within months, the company had reached out to nonprofits and landed contracts to do exactly that work.
“It was a crazy feeling,” the engineer says now. “So empowering. I felt like maybe it was possible to do good within a VC startup. Within this capitalist box, I felt like I could make a difference.”
That was then.
Tell me about how you ended up in a startup.
I grew up in a household with a dad who was an engineer, so from a young age I was nudged in that direction — with a very technocratic philosophy. By the time I got to college, it felt very natural to study computer science. My first job out of school was doing chip design at a startup that was really old, very traditional, very boring. They ran out of money and didn’t pay us, which actually worked in my favor as an international student, since I had no legal obligation to stay.
Then I found my current job, at a more hip, VC-backed startup in SF. We conduct LLM evaluations to measure AI capabilities in economically valuable domains. Finance, math, things like that.
The environment was completely different. At the last job, I felt like a cog—fungible, like nothing I did actually mattered. At the new place, everyone was young. Everyone’s basically my age. In their 20s and 30s. There’s no established way of doing things, so there’s enormous freedom to take initiative. No structure at all, and I thought that was enticing.
One of my coworkers described what we do in a way that really stuck with me: evals are a way to create a communication bridge between AI labs and communities. I thought that framing could apply to social work too – that the same methodology could speak for communities who aren’t being heard.
That was the seed of the 10 pm idea.
And then things changed.
At the start of this year, I started feeling more disillusioned. My boss began to consider working with various government agencies. He said that if we weren’t comfortable working with said agencies, we could voice that. But it felt more like a platitude. It seemed like too lucrative an opportunity for him to actually turn down.
That became the first source of my disillusionment. I realized: I actually don’t have the power to prevent that from happening. Our main goal is to grow as fast as possible. We’re supposed to be a source of truth about the state and capabilities of AI, and that means taking as many opportunities as you can. I thought if I stayed at the company, I could start my own initiatives — but the structure doesn’t really allow for that.
At the same time, I was taking a class on labor organizing in tech. The power disparity started to feel very real, and I felt like I was losing my sense of agency.
A former coworker who left his previous company as a founding engineer left because the CEO signed a deal he couldn’t ethically get behind. Now he’s at my startup, and he doesn’t clock that same connection.
That’s the thing. You watched one startup transform into this entity you didn’t recognize. Then you move to a new one and don’t realize that the same structure is being created again. Until you feel the pain of it yourself, you’re blind to it.
I think the structure of a company — if it’s not worker-first — is destined to create hierarchy. You have your boss on top, and they call the shots. Any company structured that way is destined to end up there.
Why is it so hard to act on that? What makes it hard for people there to speak up, compared to, say, working in big tech?
There are two things.
First, the scale. In a small startup, hierarchies haven’t fully emerged yet. There’s no class struggle — you can’t feel class struggle if it hasn’t materialized yet. The traditional framing of the managerial class pitted against workers, the boss as villain, the union as protection, is hard to live by when your boss is genuinely looking out for you. My boss has been supporting me through my visa process. He’s genuinely concerned about my well-being. It’s hard to pick up the pitchforks without soiling those relationships.
Second, your well-being is very closely tied to your employer. For example, being an international worker does this. It’s really hard to stay in the country without that job. All of those things together make it hard to organize at a small scale — being an immigrant, having bosses who feel like friends, not yet feeling the class struggle, even though you can sense it emerging. If I weren’t an immigrant, I’d probably have more courage. I could just find a different job.
The hard part is that the incentives are backward. No one thinks they need protection yet. But by the time the power dynamic shifts, the culture is already set.
From what you’ve observed, among your peers, in your circles, are people in startups interested in speaking out about emerging risks and harms? What have you noticed? If not, what do you think gets in the way? What about the people around you? Are they concerned?
I’m in a startup with under 20 people, so the sample size is very small. But to me, it seems like there are people who are somewhat socially conscious. I have one coworker who immediately voiced his discomfort about working with the CIA, but that’s like 1 out of 16 workers. The rest don’t really see a lot of issues with the status quo. For example, one of our interview questions was “who’s someone you admire?” and like three people said Elon Musk. They see all the problematic stuff he does as quirky and still admire his engineering approach and leadership. Also, the gender ratio at my company is pretty skewed. There are currently only three women. I do have one coworker who’s really into social benchmarks; we bonded over making social evals. She’s really interested in social policy and AI for moderation and public policy. But the majority don’t really care.
How about tech workers you know outside your company?
My friends outside the company aren’t much different. One just clocks in and clocks out. I’m sure the company is somehow connected to the military, but for him, work is just a means of sustaining himself. I have another friend who knows their work isn’t particularly meaningful, and it seems like they hold out hope that one day they’ll quit and do something more impactful. But there’s no initiative to change anything from the inside. The attitude is more like: I’m going to hop off this ship at some point and do something more meaningful. You pick the least bad tech company you can find and optimize for doing the least harm. And yet, a lot of these people are very politically active outside of work — going to protests, community farming, etc.— but when it comes to their actual jobs, they just clock in and clock out.
Do you think there’s a kind of idealism among tech workers in the Valley — especially people just starting out — that gets quietly worn down as start-ups adhere to market pressures over time? You mentioned the startup narrative specifically. What does that do to people?
CEOs are so good at pitching. They construct these narratives and create meaning out of things that may not inherently have meaning. When you’re young and hungry for purpose, you subscribe to that narrative. It also feeds the market — you work 60-hour weeks because you genuinely believe in the mission.
There are people who treat work purely as a means to sustain their lives. They tend to be drawn to big tech — coasting, going through the motions, keeping their real life outside of work. But startups market themselves to people who want to build things. You come out of college with all this energy, and you’re naturally drawn to that. You start something scrappy and change the world. The YC mantra, you know. Narratives help us understand reality, but they also shape it. Young people are locked in, willing to work 12-hour days. I’m no different. This week alone has been brutal.
Do you think there’s a difference in how those in Silicon Valley think versus those in the general public? Is there a culture in Silicon Valley that makes this harder to see from the inside?
At least in my circles, tech people are almost uniquely optimistic about the future, but it’s a kind of blind optimism. The attitude is: technology will progress, and things will land in the most optimal way. No one’s pressed.
You go to non-tech events, and people are pissed. People outside of tech don’t trust Waymos. Whereas, in tech, you don’t think twice about ordering DoorDash. One of my coworkers said it makes more sense to him to buy a new vacuum cleaner than to clean out his old one. People in tech tend to optimize for individual comfort; it’s a very individualistic culture, I think.
There’s also this hardcore, very male idealism — an obsession with conquering, with overcoming the limitations of the body. Lots of people in tech love The Fountainhead by Ayn Rand. You read it, and you really feel the power of the individual — there’s something seductive about the idea of being someone who can build a skyscraper from pure will.
And then there’s also this China scare. Companies are getting mobilized to develop technology out of a fear of falling behind China.
The general take in the industry is that AI feels like the messiah idea right now — it will either lead us to salvation or to our demise. I don’t have a complete answer to this question, but those feel like the two biggest forces shaping the culture right now.
Is there anything structural that would actually enable workers to speak out?
One thing that strikes me as a genuine model from the VC space is the SAFE contract – the standard two-page agreement that protects founders. In the music industry, artists used to sign deals that stripped them of their masters. VC kind of solved something analogous to this simple document that lays out clear terms. It would be interesting if a similarly standardized contract could have worker protections built into it from the start — before the culture sets in, before the hierarchy forms, when there’s still room to negotiate.
Because once a startup takes money with strings attached, that dictates everything. The whole VC business model makes it nearly impossible to operate freely. The north star is maximizing shareholder value. Timescales can be stretched or condensed, but that’s always the guiding principle.
The irony is that the people most motivated to change things are the ones who sign up for the startups. And the startups are the places least equipped to change.
Do you think start-up culture is different from that in big AI/tech companies? How, if at all, are the workers different?
Startup people aren’t trying to coast. They’re motivated by a bigger mission; obviously, some are money-maximizing, but I think most genuinely want to make the world a better place. These people aren’t evil. I genuinely think someone like Alex Karp believes in his mission. He thinks Western hegemony is the most optimal arrangement for the world, and if you follow the logic, it isn’t totally irrational — it just stems from a core belief and line of reasoning that most people don’t subscribe to. If you sincerely believe AI is going to make the world a better place, joining an AI startup makes total sense. That’s the box they’re operating in.
The views expressed in this interview are those of our guest. They do not necessarily represent Psst.org’s positions.
Psst.org is a nonprofit that helps tech workers disclose public-interest information safely. If you’ve seen something at work that concerns you, request a pro bono consultation with one of our lawyers via our Safe. All conversations are confidential.