Anant Agarwal’s New Act: From Scaling Access to Scaling Expert Feedback

MILLBURN, N.J. – June 30, 2026 – The appointment of Anant Agarwal as the new CEO of Grady, an AI-powered grading platform, is far more than a typical leadership announcement. It’s a powerful signal from one of the most influential figures in modern education. Agarwal, the founder of edX and a long-time MIT professor, spent over a decade evangelizing the power of Massive Open Online Courses (MOOCs) to democratize access to learning. Now, he’s placing his bet on a small, faculty-built company aimed at solving a far more granular, yet arguably more fundamental, problem: scaling expert feedback.

For leaders watching the cross-currents of technology and industry, this move warrants close attention. It represents a strategic pivot from a macro problem (access) to a micro-execution challenge (assessment quality). Agarwal's decision to lead Grady suggests a belief that the next great disruption in education won't come from broadcasting more content to more people, but from fundamentally improving the learning loop for every student, enabled by AI that augments, rather than replaces, human expertise.

The Pioneer's New Gambit

Anant Agarwal's career has been defined by a singular mission: widening access to education. With edX, he and his colleagues at MIT and Harvard broke open the doors of elite institutions, offering courses to tens of millions globally. The platform’s acquisition by 2U for $800 million in 2021 cemented the MOOC model as a permanent fixture in the educational landscape. Yet, even at the peak of this success, a persistent challenge remained. As Agarwal himself noted, "I have spent much of my career trying to democratize education. Assessment and feedback are core components of learning, and we have never been able to scale them without giving something up."

This admission cuts to the heart of his move to Grady. The MOOC era, for all its triumphs in access, often relied on imperfect solutions for assessment, such as peer grading or automated multiple-choice quizzes. These methods sacrificed the nuance and authority of expert feedback for the sake of scalability. Agarwal is now tackling this compromise head-on. His transition from Chief Academic Officer at 2U to CEO of an early-stage startup is a calculated bet that AI can finally resolve this tension. He remains a professor at MIT, grounding his new corporate role in the day-to-day realities of academic life he has known for nearly four decades. This isn't a tech executive parachuting into education; it's a seasoned educator grabbing the controls of a new, powerful tool.

An 'Exoskeleton' for the Modern Classroom

What makes Grady the object of Agarwal's bet is its core design philosophy. Founded by two university professors and AI researchers, Periklis Papakonstantinou and Anastasios Sidiropoulos, the platform was engineered from the instructor's perspective. It is not, its creators insist, an "autonomous robot" designed to automate educators out of a job. Instead, they use a more compelling metaphor: an "exoskeleton that amplifies the human instructor."

In practical terms, this means the instructor remains firmly in command. They set the grading rubric and define the standards for an assignment. Grady’s AI then acts as a force multiplier, applying that rubric with perfect consistency across hundreds of submissions. It can identify patterns, group similar responses, and draft detailed feedback based on the instructor’s criteria. Crucially, the instructor reviews, edits, and approves everything before it reaches the student. This "human-in-the-loop" model is a deliberate design choice, one that aims to build trust in a sector deeply skeptical of technological overreach. "It holds every student to the instructor's standard, and it keeps the instructor in charge," Agarwal explained. "It effectively supercharges faculty and teaching assistants." This approach directly addresses the "hidden challenge" of execution in AI: ensuring the technology serves the expert user, not the other way around.

Solving the Scalability-Quality Paradox

The problem Grady addresses is one familiar to any large-scale organization, not just universities. How do you maintain quality and consistency when an operation grows? In higher education, a seminar with ten students allows for rich, personalized feedback. In a 400-person introductory course, that ideal shatters. The work is typically divided among a team of teaching assistants, each with their own interpretations and levels of fatigue. The result is human variability, where the same work could receive a different grade and different comments depending on who grades it.

Grady promises a quantifiable benefit: radical consistency. Whether an assignment is the first or the 800th to be graded, the AI applies the instructor's standard without fatigue or bias drift. This not only ensures fairness but also generates valuable data. By aggregating its findings, the platform can provide instructors with a clear, real-time dashboard of which concepts the class has mastered and where they are struggling. This transforms grading from a purely summative, after-the-fact exercise into a formative tool that can shape teaching while the course is still in progress. Students, in turn, receive detailed feedback while the material is still fresh in their minds, a critical factor for effective learning that is often lost when grading takes weeks.

Navigating the AI Trust Deficit

Agarwal is entering a crowded and contentious field. AI in education is fraught with concerns about algorithmic bias, data privacy, and the erosion of the human connection central to learning. Competitors like Turnitin and Gradescope are already well-entrenched, focusing on everything from plagiarism detection to workflow efficiency. Grady's strategic differentiator is its explicit focus on earning trust. As Swaroop Kolluri, Managing Director at investor Neotribe Ventures, put it, "We look for AI that earns trust where trust is hardest to earn."

By positioning itself as a faculty-centric "exoskeleton," Grady directly confronts the primary fear that AI will de-skill and displace educators. The company's narrative is one of empowerment, not automation. This is a savvy market position. In an era where generative AI can produce student essays and faculty are on the front lines of discerning authentic work, a tool that promises to enhance, rather than replace, their judgment is likely to find a more receptive audience. Agarwal’s reputation as an advocate for education lends significant credibility to this vision. His leadership signals that Grady is not just another tech-first solution looking for a problem, but a pedagogy-first tool designed to solve one of education's most persistent operational bottlenecks. The implications extend beyond the university; it is a case study in how to deploy AI in expert-driven fields where judgment, nuance, and trust are paramount.