How AI labs use Mercor to get the data companies won’t share

AI Labs Turning to Former Industry Experts for Data Insights

Instead of entering into costly contracts with companies for access to their data, AI labs are adopting a new approach by leveraging the knowledge of former senior employees from those companies. This revelation was shared by Mercor CEO Brendan Foody during the JS Disrupt 2025 event on Tuesday.

Foody highlighted Mercor’s marketplace as a platform connecting ex-employees of investment banks, consulting firms, and law practices with AI labs seeking to automate processes within those industries. Some notable clients of Mercor include OpenAI, Anthropic, and Meta.

Discussing the significance of this strategy, Foody mentioned that companies like Goldman Sachs may not be enthusiastic about the idea of models automating their value chain. This shift in dynamics has led AI labs to engage contractors with prior industry experience to train models for automation, a service that Mercor provides.

At just 22 years old, Foody co-founded Mercor, which compensates industry experts up to $200 per hour for tasks like form filling and report writing for AI training purposes. The startup boasts tens of thousands of contractors and pays out over $1.5 million daily. Despite these expenses, Mercor remains profitable due to the high value placed on the data provided by these experts.

Within three years of its establishment, Mercor has seen its annual recurring revenue soar to approximately $500 million, recently securing funding at a valuation of $10 billion.

However, the rise of Mercor has raised concerns among established players in various industries, as the knowledge of their operations could potentially be leaked through former employees on the platform, leading to the automation of their work processes. Foody acknowledged this apprehension but refrained from labeling it as a loophole.

Embracing the evolving landscape of work, Foody suggested that Mercor’s marketplace could pave the way for a new gig economy similar to Uber’s impact in the past. Notably, Uber’s former chief product officer, Sundeep Jain, has joined Mercor as president.

While Mercor aims to extract knowledge from diverse industries, Foody emphasized the importance of preventing corporate espionage among contractors, who predominantly hail from secretive sectors like law and finance. Despite instructions against uploading documents from their former workplaces, Foody acknowledged the challenges posed by the scale of the operation.

Arguing that an employee’s knowledge belongs to the individual rather than the company, Mercor sometimes treads a fine line in its job postings by seeking access to proprietary data for AI evaluations or model training.

Mercor was among the pioneering data startups in the U.S. to recruit skilled knowledge workers and offer substantial compensation for AI model training. While early players like Scale AI initially engaged contractors from developing countries for basic labeling tasks, competitors like Surge and Scale AI have since recognized the value of industry experts in enhancing AI models.

Despite facing setbacks, including a decline in collaborations following a significant investment from Meta in Scale AI, Mercor has experienced rapid growth, albeit remaining smaller than its competitors valued at over $20 billion.

Looking ahead, Foody envisions expanding Mercor’s partnerships beyond AI labs to industries like law, finance, and healthcare, aiming to assist these sectors in leveraging data for AI training purposes.

“Over time, ChatGPT will be better than the best consulting firm, better than the best investment bank, and better than the best law firm,” Foody predicts. “This transformation will have a positive impact on the economy, fostering abundance for all stakeholders.”