Valon Makes Most New Hires Earn AI Access, Projects 75% Cut in Token Spending

Valon now requires most new hires, including senior recruits in nearly every business function, to learn their roles without AI. Managers must decide that a worker can identify incorrect AI output before granting access.

CEO and cofounder Andrew Wang introduced the policy after reviewing employee use of broadly available AI tools. Wang said workers had turned to the most expensive models for simple assignments because they believed the systems were usually correct; he also said experienced employees had to clean up poor AI-generated work from newer hires.

The restriction is a notable constraint for a company with roughly 320 employees that builds mortgage-servicing software powered by AI agents. It applies across nearly all functions rather than only to junior staff.

• Most new hires begin without AI, including senior recruits.
• Managers decide when a hire understands the role well enough to recognize erroneous output.
• Engineers are exempt because Valon requires peer review of all code before release. Wang said finance and human-resources teams lack comparable safeguards.

Wang said the rule is already directing new recruits to more tenured colleagues for help instead of asking an AI system to solve problems. He said those conversations are giving recruits a deeper understanding of their work. The savings estimate and the learning benefit are both Wang’s assessments of the policy’s early effect.

A September 2025 BetterUp survey conducted with Stanford’s Social Media Lab found that 40% of 1,150 full-time U.S. desk workers had received AI-generated work from a colleague in the prior month. Respondents said dealing with each instance took nearly two hours on average.

Wang said some AI enthusiasts criticized the policy in comments on a LinkedIn post. He also said experienced Valon employees generally welcomed it because they had been correcting poor-quality work from newer hires. Wang’s response to critics was an invitation to offer a better way to ensure people learn the underlying work.