Better Data, Better AI | imerit

The Shift Towards High-Quality Data in AI Development

iMerit, an AI data platform, posits that the progression of AI integration within enterprises hinges not on an increased volume of data, but on enhanced data quality. This superior data, according to the company, is not sourced from extensive networks of freelance workers, but from specialists possessing expertise in fields like mathematics, medicine, finance, and autonomous systems.

The Importance of Cognitive Expertise

“The capacity to attract and retain top-tier cognitive experts is now paramount,” stated Radha Basu, CEO and founder of iMerit, in an interview with TechCrunch. “These experts are crucial for customizing large models to effectively address specific enterprise AI challenges.”

For nearly a decade, iMerit has operated as a reliable data annotation partner for organizations involved in computer vision, medical imaging, autonomous vehicle development, and other AI applications demanding precise, human-verified labeling.

Introducing the Scholars Program

iMerit is now officially launching its Scholars program, having completed its beta phase. The program’s primary objective is to cultivate a growing team of experts dedicated to refining generative AI models for enterprise use, and increasingly, foundational models.

The company currently serves a significant clientele, including three of the leading seven generative AI companies, eight prominent autonomous vehicle firms, three major U.S. government agencies, and two of the top three cloud service providers.

Market Dynamics and Competitive Positioning

This announcement arrives amidst changes at Scale AI, a major competitor in AI data annotation, following the departure of its founder and CEO, Alexandr Wang, to Meta, along with a 49% acquisition by Meta. Several of Scale AI’s key clients, including Google, OpenAI, Microsoft, and xAI, subsequently reduced their engagement due to concerns about potential access to their product strategies by Meta.

iMerit differentiates itself by not attempting to replicate Scale AI’s focus on high-volume, rapid “blitz data.” Instead, the company is concentrating on expert-driven, high-quality data—data that necessitates nuanced human judgment and specialized domain knowledge.

A Focus on Quality and Expertise

“We represent a more mature approach,” explained Rob Laing, iMerit’s VP of global specialist workforce, to TechCrunch. “Substantial investment is currently flowing into AI. While many are building extensive human workforces, the resulting output often lacks the quality required by enterprises.”

Basu illustrated this point with the example of healthcare scribes utilizing foundational large language models. Without the input of medical professionals, the accuracy of such systems may only reach 50% or 60%.

“The goal is to achieve 99% accuracy,” Basu emphasized. “Expert-led AI enables the questioning, testing, and refinement of models for enterprise applications.”

The Ango Hub Platform and Expert Retention

iMerit’s specialists utilize the company’s proprietary Ango Hub platform to fine-tune both enterprise and foundational AI models. This platform allows “Scholars” to interact with client models, generating and assessing challenges to improve performance.

Attracting and retaining these cognitive experts is central to iMerit’s strategy, as these professionals typically engage in long-term projects spanning multiple years. The company reports a 91% retention rate, with women comprising 50% of its expert workforce.

Building a Collaborative Community

Drawing on his experience with the human translation platform myGengo, Laing noted that securing a large workforce is relatively straightforward. However, fostering a thriving community requires a more people-centric approach.

“New Scholars are introduced to the team and participate in collaborative discussions,” Laing stated. “They are encouraged to operate at the highest possible level, and our selection process is exceptionally rigorous.”

“We anticipate that companies prioritizing engagement, retention, and quality, like iMerit, will become the preferred partners for AI training in the coming years,” Laing added.

Scalability and Financial Sustainability

iMerit currently collaborates with over 4,000 Scholars and intends to expand this number. Despite not having raised additional funding since 2020—when it secured investments from Khosla Ventures, Omidyar Network, Dell.org, and British International Investment—iMerit is both sustainable and profitable.

Basu indicated that the company’s existing cash reserves allow it to scale to 10,000 experts. Further expansion would necessitate external investment, but iMerit is not currently pursuing it aggressively.

Future Growth and the Demand for Specialized Data

iMerit has been developing the Scholars program for the past year, initially focusing on healthcare. The company plans to extend its reach into other enterprise applications, including finance and medicine. Laing highlighted generative AI as its fastest-growing sector, with leading AI firms leveraging iMerit to enhance their foundational models.

“The readily available data on the internet is diminishing, and low-input human data is becoming increasingly commoditized,” Laing concluded. “Organizations are now seeking to refine these models to achieve artificial general intelligence (AGI) or superintelligence.”