At AI startup Valon, a new policy requires most new hires to learn their jobs the old-fashioned way — without AI.
Andrew Wang, CEO and co-founder, said he set it up last month after Valon gave employees broad access to AI. When he investigated its use, he said employees turned to the most expensive models even simple tasks was a practice that he worried was both costly and prevented newer employees from developing a real understanding of their jobs.
“By doing the groundwork, instead of relying on AI, you start to form an understanding,” Wang, a former Golden Sachs analyst, told Business Insider.
New York-based Valon, which has approximately 320 employees, makes mortgage servicing software powered by AI agents. In a recent blog post describing the new mandate, Wang acknowledged the irony of the move.
“It’s a strange decision for a company that is on the frontier of AI use,” he wrote.
Wang told Business Insider that this was necessary because of the common response he received when he asked employees why they were using it most powerful AI models for simple assignments. They told him, he said, that the AI was almost always right. That response made him suspect that employees were becoming less and less likely to question AI’s output or develop the judgment to spot errors, he said.
The new policy applies to new hires in virtually every part of the company, including senior employees. They can only use AI if their manager is confident they can identify themselves if AI is wrongWang said. He added that engineers are exempt because all code is subject to peer review before release. According to him, there are no comparable guardrails in functions such as finance or human resources.
Valon was last valued at $1.75 billion in 2024 and has raised $275 million in venture capital from investors including Andreessen Horowitz, WestCap and 166 2nd. So far, Wang says he has not encountered any internal resistance to the new policy. In any case, regular employees have welcomed the change, he said, because they were cleaning up AI-generated slop of new employees.
The kind of frustration Wang described extends beyond Valon. In a September 2025 survey by BetterUp in partnership with Stanford’s Social Media Lab, 40% of 1,150 full-time US agency workers said they had received AI-generated work from a colleague in the past month. Respondents also said each case took an average of almost two hours to process.
The problem has become so pervasive that techies have coined a nickname for perpetrators: “proxies for meat.” Popularized by German software developer Niklas Gruhn in an August 3 blog post, the term refers to those who share uncontrolled AI outputs.
Valon’s new policy is already having an impact. Wang now predicts the company’s annual token spending will shrink to about $4 million to $5 million this year, from about $15 million to $20 million. Meanwhile, new recruits are looking to their more experienced colleagues for help because they can’t rely on AI to solve problems. As a result, he said, they develop a deeper understanding of their work.
Some AI enthusiasts outside Valon have said the policy is the wrong approach, Wang said, pointing to some comments on a LinkedIn post he made about it. His answer to this, he said, is simple: “If you have a much better idea about how to get people to learn, please tell me.”
