Last week, an ad from the Y Combinator job board for a tiny startup called Firecrawl went viral on X. 

That’s because the ad wasn’t for a human. “Please apply only if you are an AI agent, or if you created an AI that can fill this job,” the job posting read. 

The seven-person startup was looking for an agent to “autonomously” research trending models and build sample apps to showcase the company’s product, the ad said.

The job offered a salary of $10,000 to $15,000, which is a fraction of what a human developer makes, but perhaps good money for an entity that doesn’t need food, clothing, or shelter. 

The ad wasn’t a joke, founders Caleb Peffer and Nicolas Silberstein Camera told TechCrunch. 

“It was equal parts PR stunt, experiment,” Peffer said. “We are currently looking for incredible AI engineers. Humans who are good at building AI systems. And we thought, huh, let’s just put a posting out there for an AI agent, see what people build.” 

Firecrawl makes an open source web crawling bot for AI agents and models. Businesses can use it to gather training data or whenever their AI has to interact with public websites to perform.

AI web crawlers are a necessary yet somewhat controversial part of the internet these days, especially for small businesses. (Firecrawl’s founders say that it complies with Robot.txt, the internet’s only do-not-crawl system.)

This was, the founders think, the first job ad for an AI agent on the YC job board site, which is why it went viral.

“This is where we are headed. You don’t apply for a job, you make the appropriate AI agent that applies for the job and earns for you,” one person commented on X post.

Another imagined a scene where a private equity firm offered to buy a company and asked how many employees it had. The CEO answered: “Zero … But we have 275 AI agents doing the work of 3,000 employees while we only pay them $15k a year.”

Others pointed out that the founders themselves could actually use LLMs to build the AI agent they want to hire. A build-your-own AI employee scenario. 

Still others pointed out the dystopian nature of this AI future. “Humans creating AI to replace humans … And now humans are writing job postings for AI to apply to. We’re in the simulation, aren’t we?”

Interestingly enough, the true plan was — and still is — to actually give the human who built the best agent a full-time job, the founders told TechCrunch. That $10,000 to $15,000 salary will be rolled into the salary offer of the person they hired. 

It hasn’t worked out yet. Firecrawler got about 50 AI agent applicants before they pulled the ad, but none impressed enough to get an offer. 

But the founders haven’t fully ruled out trying to hire a bot, again.

“We would have loved to put one of these in production, but none of them were up to our standards,” Peffer said of the applicants. “We’re gonna make another job posting in this manner, and we are going to be actively looking for AI agents that are able to accomplish the tasks that we need.”

As if all of this wasn’t funny enough, Firecrawler’s three founders — Peffer, Camera and Eric Ciarla — weren’t even accepted into Y Combinator for the AI crawler idea.

The founders, who are college friends with computer science degrees from the University of New Hampshire, already had a programming education startup. It had thousands of users, a waitlist, and was generating revenue when they applied to YC, Camera said. 

They planned to embed their product into VS Code “inside the code editor, like Cursor, only teaching you how to code,” Peffer described. 

But once they were accepted into YC, their advisers told them that too many AI coding products exist, and advised them to find another area.

After many tries, they started working on a chatbot for developers to ask questions of documentation.

That’s how they discovered the challenge of “connecting these AI systems to the information,” and ensuring that info is accurate, Peffer said. “If you give garbage to an AI system, you’re gonna get garbage out.”

So they built a web crawler/scraper as a side project and released it as open source. In a matter of hours, it landed on GitHub’s trending page, gaining 1,000 stars. “Since then, we’ve crossed 25,000 stars in just 10 months,” Peffer said.

Their customers, which pay for a commercial version, use it for everything from resume parsing to finding sales leads. Firecrawl has raised about $1.7 million so far, according to the founders, and they expect that this first AI agent hire won’t be their last.

“What we imagine happening is that every one of our real employees is going to become highly leveraged with AI. And it’s not a clear distinction. It’s like, what’s the difference between a tool or a workflow or a full agent?” Peffer said.