EXL Acquired the Company That Trains AI Models. Now It Is Selling the Full Stack.

ExlService Holdings announced on August 3, 2026 the completion of its acquisition of iMerit, a company that works at the layer of AI development most enterprises never see but cannot do without: the training, evaluation, and alignment of the models that power their AI systems.

The acquisition adds model development capabilities to EXL's existing AI and analytics business, which focuses on regulated industries including insurance, healthcare, banking, retail, communications, and energy. 

Before the acquisition, EXL specialized in applying AI within industry workflows. iMerit adds expertise in training, fine-tuning, and evaluating the underlying AI models used in those environments, the company said.

"This acquisition is a transformational pivot for EXL, deepening our vertically specialized AI capabilities and expanding our reach into high-growth AI technology sectors," said Rohit Kapoor, Chairman and CEO of EXL. "We are well positioned to help clients build, fine-tune and operationalize AI that performs reliably in production, especially in regulated industries where domain knowledge, context and compliance are non-negotiable."

iMerit operates at three levels that are distinct from EXL's existing capabilities. Scholars is a global network of domain specialists like physicians, scientists, engineers, and linguists who work on data creation and model alignment for AI labs and enterprises. 

The specialist composition of that network matters: training a medical AI system requires different human expertise than training a legal one, and iMerit's workforce is structured around domain specificity rather than generalist annotation.

Ango Hub is iMerit's multimodal data collaboration platform, the technical infrastructure through which that human expertise is organized and applied to model training workflows. 

And iMerit's existing relationships with foundation model labs, built through fine-tuning and evaluation work, give EXL direct access to the AI infrastructure layer at a time when those relationships are among the most commercially valuable in enterprise technology, according to the press release.

"The next generation of enterprise AI will be defined not by the models organizations choose, but by how effectively they can deploy them in real-world business environments," said Radha Ramaswami Basu, joining EXL as Executive Vice President, Head of iMerit, and a member of the executive committee. "Together, we can help clients bridge the gap between innovation and execution, turning AI potential into measurable business results."

The framing connects directly to the deployment gap documented consistently this year; research from Forrester, Anaconda, and IDC has found that roughly 88% of enterprise AI pilots never reach production. 

EXL's argument is that the combination of model-layer capability and domain expertise in regulated industries is the specific combination required to close that gap.