Mercor Success Story: How Brendan Foody Built a Business Recruiting Experts to Train AI
In 2023, Brendan Foody and two former schoolmates launched Mercor with a straightforward ambition: make hiring faster and more effective using artificial intelligence. Within three years, their company had evolved into a major supplier of specialist human expertise for the AI industry, reached a valuation of $10 billion and established relationships with some of the world’s leading AI developers. Mercor’s success was not simply a matter of entering a fashionable market at the right time. Foody and his co-founders recognised that increasingly sophisticated AI systems still needed experienced people to teach them how professional work is actually done. By turning that requirement into a commercial service, they created a rapidly growing business connecting doctors, lawyers, engineers, financial specialists and other professionals with organisations developing advanced AI. The story of Mercor illustrates how a young company can change direction, identify an overlooked commercial opportunity and build a valuable business around a problem that larger organisations have yet to solve.
How Brendan Foody Founded Mercor and Identified a Market Opportunity
Brendan Foody’s entrepreneurial career began unusually early. Before founding Mercor, he attended Bellarmine College Preparatory in California, where he met Adarsh Hiremath and Surya Midha through competitive speech and debate. Their shared interest in argument, research and problem-solving helped establish a working relationship long before they became business partners. Foody subsequently attended Georgetown University, while Hiremath and Midha pursued studies at Harvard. None of the three had extensive corporate experience or a conventional recruitment background. Nevertheless, they believed that modern technology could address a problem familiar to employers everywhere: finding people with the right abilities was often unnecessarily slow, expensive and inconsistent. Rather than waiting until they had completed university, the friends decided to test their idea while still in their teens.
Mercor was founded in January 2023, with Foody taking responsibility as chief executive, Hiremath overseeing technology and Midha managing operations. Their initial proposition was not specifically about training AI models. They wanted to improve the recruitment process itself by using AI to assess applicants and match them with suitable opportunities. Traditional recruitment frequently involves reviewing hundreds of CVs, arranging preliminary interviews and comparing applicants whose qualifications may not accurately reflect their abilities. Mercor introduced automated assessments and AI-led video interviews that could help employers identify promising candidates without manually screening every application. The system could examine professional experience, technical skills and interview responses, giving companies a more structured way to evaluate applicants. Although automated interviews cannot eliminate hiring bias or guarantee accurate judgments, they offered a potential way to reduce the administrative burden of recruitment.
The founders initially worked on their business from university accommodation, relying on their own resources before securing substantial outside investment. By January 2024, when Mercor publicly announced its $3.6 million seed financing led by General Catalyst, the company reported a talent network of approximately 100,000 people across 25 countries and annual recurring revenue in seven figures. Those were company-reported figures rather than independently audited results, but they indicated that the concept had attracted both applicants and paying customers. The early funding supported further development of its recruitment technology and helped Mercor expand beyond the small businesses it initially served. More importantly, the first year gave Foody direct experience of an important commercial reality: companies were willing to pay for access to carefully selected talent, particularly when specialist knowledge was difficult to find through conventional recruitment channels.
From Student Entrepreneurs to a Venture-Capital-Backed Company
Mercor’s early development depended on a combination of complementary skills and a willingness to challenge established recruitment practices. Foody concentrated on business development, partnerships and the company’s overall direction, while Hiremath focused on the technology required to assess large numbers of applicants. Midha handled operational responsibilities, including the systems needed to support employers and contractors. Their experience in competitive debate also proved useful because building a young business required them to defend assumptions, respond to criticism and communicate complex ideas to potential customers and investors. All three eventually left university to work on Mercor full-time and became Thiel Fellows, joining a programme associated with entrepreneur Peter Thiel that supports young people pursuing ambitious projects outside conventional higher education.
Securing investment helped turn Mercor from a promising experiment into a business capable of serving larger customers. In September 2024, the company announced a $32 million Series A funding round led by Benchmark, with a reported valuation of approximately $250 million. The investment represented a significant endorsement of its approach to recruitment at a time when employers were increasingly interested in AI-assisted hiring. Yet the money was only part of the equation. Mercor also needed a dependable supply of qualified candidates, efficient screening procedures and customers willing to trust a relatively new company with important recruitment decisions. Maintaining the quality of its talent network became particularly important as the volume of applications grew. A recruitment service can attract large numbers of candidates relatively quickly, but its lasting commercial value depends on whether employers consistently receive people who can perform the work.
A major advantage emerged from the way Mercor approached matching workers with opportunities. Rather than concentrating exclusively on permanent employment, the company also connected organisations with professionals available for project-based assignments. This made the business particularly relevant to AI developers, whose requirements could change quickly as new models were being tested and improved. A laboratory might need mathematicians for one research project, software developers for another and legal specialists for a third. Recruiting each group through a traditional hiring process could take weeks or months, while maintaining all those specialists as permanent employees might be financially impractical. Mercor’s experience in identifying, assessing and coordinating contractors positioned it to respond to this demand. What began as an attempt to improve recruitment was gradually becoming a way to organise specialist human knowledge for the AI industry.
How Mercor Turned Expert Recruitment Into a Multibillion-Dollar Business
Mercor’s decisive commercial opportunity came from a growing limitation in artificial intelligence. Although advanced AI models could already generate convincing text, write computer code and answer technical questions, they frequently struggled with the judgement required in specialist occupations. Medical reasoning, legal analysis, financial modelling and engineering decisions involve more than reproducing information from textbooks or websites. Professionals must interpret incomplete evidence, recognise exceptions, evaluate competing options and understand the consequences of their recommendations. AI developers therefore needed more than enormous collections of general information. They needed carefully prepared examples, assessments and feedback from people who understood these occupations. Mercor recognised that its existing recruitment capabilities could help AI companies find such specialists at scale, and the demand for this work rapidly became central to the business.
Under this model, Mercor recruits professionals whose knowledge can contribute to the development and evaluation of AI systems. A lawyer might review an AI-generated contract analysis and identify overlooked legal considerations. An accountant could assess whether a financial report follows recognised professional procedures, while a software engineer might test whether an AI assistant has correctly identified a programming error. Doctors and medical specialists can help evaluate responses to clinical scenarios, although such assessments do not replace the standards required for real patient care. The experts may also create realistic assignments, compare alternative answers and explain why one approach is more appropriate than another. Their feedback gives AI developers examples of the reasoning and practical standards used by experienced professionals. Mercor’s commercial role is to identify suitable contributors, assess their qualifications, coordinate their work and help customers obtain usable expert input.
The distinction between ordinary data collection and specialist evaluation is fundamental to understanding Mercor’s business. General information is widely available, but reliable professional judgement is much harder to reproduce. A collection of financial documents, for example, does not necessarily explain how an investment banker evaluates a transaction when the available evidence is incomplete. An experienced banker can provide that additional context by showing which assumptions matter, where errors typically arise and how a finished analysis should be assessed. AI developers can use this material to improve and evaluate their systems, although better training examples do not guarantee that a model will always make the right decision. Mercor effectively turned professional experience into a service that could be delivered across multiple research projects. As developers sought increasingly specialised forms of feedback, this created demand for workers from a wide range of industries rather than only the technical professions traditionally associated with AI.
Funding, Revenue Growth and Mercor’s $10 Billion Valuation
Mercor’s financial expansion accelerated dramatically during 2025. In February, the company announced a $100 million Series B financing round led by Felicis, valuing the business at $2 billion. That represented an eightfold increase from its reported valuation following the previous major funding round. Other investors included Benchmark, General Catalyst, DST Global and Menlo Ventures. By September 2025, reporting from TechCrunch indicated that Mercor’s annualised revenue run rate had moved beyond $450 million, according to information discussed with investors. Foody said the actual figure was higher, while explaining that the company’s revenue included the full amount billed to customers before payments to contractors. This distinction matters because gross customer billings are not equivalent to the money Mercor retains after compensating its experts and covering operating costs. The impressive growth figures therefore demonstrated demand and commercial scale, but they did not establish the company’s profitability.
The next major milestone arrived on 27 October 2025, when Mercor announced a $350 million Series C funding round, again led by Felicis, with participation from Benchmark, General Catalyst and Robinhood Ventures. The transaction valued the privately held company at $10 billion, five times its valuation only eight months earlier. Across its announced seed, Series A, Series B and Series C rounds, Mercor had raised approximately $486 million. Such a valuation reflected investors’ expectations about future growth rather than the company’s cash reserves or annual profits. Nevertheless, it established Mercor as one of the most valuable young businesses supplying human expertise to AI developers. The rapid increase also brought extraordinary personal wealth on paper to its founders. In November 2025, Forbes identified Foody, Hiremath and Midha, then aged 22, among the world’s youngest self-made billionaires, based on estimates of their ownership stakes in Mercor.
By 2026, the scale of Mercor’s operations had expanded well beyond its original recruitment proposition. The company’s published figures described a network exceeding five million domain experts, with more than $4 million being paid to contributors each day. Mercor also reported a workforce of more than 400 employees across San Francisco, New York and London. These figures are useful indicators of scale, although they should be understood as company disclosures rather than independently verified measures of business performance. The company does not publicly release the same comprehensive financial statements as a listed corporation, making its profitability and precise operating margins difficult to establish. Its $10 billion Series C valuation remained the clearest publicly confirmed benchmark for the business as of October 2026. The important achievement was not simply reaching a large headline valuation, but developing a service that major AI organisations were prepared to use repeatedly for demanding specialist projects.
Mercor in 2026: Growth, Challenges and the Future of Expert-Led AI
Mercor’s direction in 2026 showed how far the business had moved from its beginnings in automated recruitment. On 21 January, the company introduced APEX-Agents, a research benchmark designed to measure whether AI assistants could complete realistic professional assignments in investment banking, management consulting and corporate law. Instead of testing whether a model could answer a single question, the assessments examined its ability to work through complicated assignments involving multiple documents, instructions and decisions. Mercor recruited experienced professionals to help design the scenarios and determine what would count as satisfactory work. Its initial research found that leading AI agents completed fewer than 25% of the tasks successfully on their first attempt, while the best results reached only 40% even when eight attempts were allowed. These findings concerned the specific models, tasks and testing conditions used in the study, but they illustrated why expert evaluation remained commercially important despite rapid improvements in AI capabilities.
Mercor continued developing this research during 2026, extending its evaluations into software engineering and accounting. In March, it introduced a software engineering assessment developed with input from Cognition, focusing on practical tasks such as identifying faults, making changes and maintaining working systems. In July, Mercor announced an accounting benchmark developed with Ramp that included 160 assignments based on ten simulated companies. By September, the company had also revised its APEX-Agents assessment to make scoring more closely reflect professional expectations. These initiatives strengthened Mercor’s position in a market where customers increasingly wanted evidence that AI could carry out valuable work, not simply produce fluent answers. They also gave the business another use for its professional network: experts could help establish the standards against which AI products were judged. This created a connection between recruitment, training, testing and commercial deployment.
Another important development came on 26 March 2026, when Mercor announced an expansion into enterprise AI services. The objective was to help businesses build and improve AI assistants that could understand internal procedures and complete organisation-specific assignments. For example, a financial services firm might want an assistant that can prepare reports using its own templates, follow established review procedures and recognise when a result requires human approval. Mercor proposed applying the same expert-led evaluation methods it had developed for AI laboratories to these everyday business requirements. In July, the company also announced its intention to acquire Deeptune, whose technology supports realistic training environments for AI agents. These moves represented a broader commercial strategy: rather than depending entirely on supplying contractors to research laboratories, Mercor was looking for additional ways to help organisations apply professional knowledge to AI development. Its long-term prospects would increasingly depend on how successfully these additional services delivered measurable value to customers.
What Mercor’s Success Reveals About Leadership and the AI Labour Market
Mercor’s growth has also brought difficult questions about security, worker protections and operational responsibility. In March 2026, the company was affected by a supply-chain cyberattack involving compromised versions of the LiteLLM software. Mercor subsequently disclosed the incident and, in June, published the findings of an investigation conducted with external security specialists. The company stated that only a limited subset of its expert network had sensitive personal information affected, although the incident generated concern among customers and contractors. Litigation filed during 2026 also raised allegations relating to the handling of personal data, automated recruitment practices and the classification of workers. Those allegations should not be treated as established legal findings. Nevertheless, the episode illustrated an important risk in Mercor’s business model: an organisation handling professional records, interviews and commercially sensitive assignments must maintain strong safeguards and retain the confidence of both its customers and contributors.
There are broader uncertainties surrounding the market Mercor serves. Its competitors include established AI data businesses such as Scale AI, Surge AI and Turing, alongside other companies offering specialist training and evaluation services. Some major AI developers may also decide to recruit experts directly or build their own internal assessment teams. Mercor must therefore demonstrate that its combination of recruitment, quality control, project management and research expertise offers advantages that customers cannot easily reproduce themselves. For the professionals undertaking the work, the opportunities can be attractive because they provide access to specialised projects outside traditional employment. However, individual earnings, assignment availability and working conditions can vary considerably. Project-based work does not necessarily provide the stability, benefits or career progression associated with permanent employment. Mercor’s continued growth will depend partly on whether it can maintain a dependable relationship with its contributors while meeting the increasingly demanding requirements of AI customers.
The central lesson from Brendan Foody’s experience is that business success often comes from recognising how an existing capability can solve a more valuable problem. Mercor began by helping employers find suitable candidates, but its founders realised that the same ability to identify and organise skilled people could address an urgent requirement in AI development. They built on their early recruitment experience, attracted investment, developed relationships with research organisations and expanded into measuring how well AI performs professional work. Foody was not solely responsible for this outcome: the contributions of Adarsh Hiremath, Surya Midha, the wider management team and Mercor’s expert network were essential to the company’s development. By October 2026, Mercor had established a substantial business around the idea that advances in artificial intelligence still depend on human knowledge. Its future remains uncertain, but its first three years demonstrate that understanding what customers genuinely need can be more valuable than simply following the most visible technology trend.