iMerit Scholars targets domain expertise gap in frontier AI

EXL, the Nasdaq-listed data and AI services group, has launched iMerit Scholars, a purpose-built platform for recruiting, managing and paying the domain experts who train, evaluate and stress-test frontier AI models. The launch arrives three months after EXL acquired iMerit in July, and marks the first substantive product move under the combined entity. With a live community of 30,000 vetted professionals drawn from 118 countries and covering 106 languages, the platform is positioned as workforce infrastructure for the next phase of generative AI, one in which raw compute power matters less than the calibre of human judgement layered on top of it.

The announcement reframes a debate that has dominated AI investment circles for two years. Frontier labs have poured capital into GPU clusters and foundation-model parameters, but a growing body of evidence suggests the binding constraint is now domain-specific human knowledge: clinicians who can catch a model's diagnostic reasoning error, mathematicians who can trace a chain-of-thought failure, agronomists who can flag a hallucinated yield projection. iMerit's platform formalises what has until now been an ad hoc market, applying structured workflows, including RLHF, red-teaming, chain-of-thought tracing and Q&A prompting, to sectors where a model error carries regulatory or clinical consequence.

From frontier labs to regulated enterprises

The strategic logic behind the EXL acquisition becomes clearer in context. EXL already serves large regulated institutions in insurance, healthcare and banking; iMerit brings the human-in-the-loop infrastructure those institutions need before they can deploy AI at scale. Radha Basu, head of iMerit at EXL, framed the pivot explicitly: "We need people who are knowledgeable enough not just to teach models, but to prove them wrong and correct them. That parallel human authority is what will help move AI from innovation to trusted execution."

The platform is powered by iMerit's proprietary Ango Hub, which enables customers and experts to collaborate on complex multimodal data and produce validated training artefacts. Global payments and compliance infrastructure are handled through a partnership with Trolley, a cross-border payouts provider, allowing iMerit to onboard and pay specialists across 118 countries without running bespoke payroll in each jurisdiction. The combination of credentialled expert matching, secure data environments and compliant global payments is designed to lower the operational friction that has historically made scaling RLHF-style programmes slow and expensive.

The wider capital and talent landscape

The convergence angle here extends well beyond a single product launch. As AI deployment pushes deeper into healthcare, financial services and physical AI, autonomous vehicles, robotics, industrial systems, the demand for domain-literate human evaluators is structurally expanding. What iMerit is building resembles, in workforce terms, what contract research organisations built for clinical trials: a scalable, credentialled, globally distributed pool of specialists that any organisation can access without internalising the recruitment and compliance overhead.

That model has clear implications for how capital flows into the AI services sector. Investors who priced the AI boom through hyperscaler capex and foundation-model valuations are beginning to look at the human-capital layer as a distinct and durable revenue stream. Services firms including Accenture, Cognizant and Appen have each staked out positions in data labelling and AI evaluation, but few have built the domain-expert credentialling infrastructure at the scale iMerit is describing. Whether EXL can sustain a differentiated position, or whether the platform becomes a commoditised marketplace as competition intensifies, is the strategic question that will define the next stage of the company's AI narrative. The regulated-enterprise focus, where trust and auditability matter more than speed, may be the most defensible ground available.