Agentic tutors: the end state of education

The Problem

Something peculiar happens when you watch a sixth grader struggle with pre-algebra. The knowledge exists: in textbooks, videos, a thousand Khan Academy lessons. Content quantity is not the problem. Yet learning fails to happen.

This failure reveals something fundamental: teaching is not just about the content. It's the art of presence—catching confusion as it forms, adjusting to a furrowed brow, intervening at the precise moment struggle threatens to become surrender.

Learning requires a tight feedback loop that most software cannot provide. Information is static. Teaching is proactive and responsive.

Why Now

We've built our schools on a flawed premise, and the results are catastrophic. Reading comprehension has collapsed to 1992 levels. Math proficiency sits at its lowest point in two decades. Only one in four students can write coherently. Critical thinking skills erode year over year.

Until now, great tutors didn't scale. They cost $100+/hour and live in major cities. But we've crossed an interesting threshold: LLMs are not only incredible content generation machines, but they can also now maintain pedagogical state across months of sessions, adapt explanations in real-time, and—critically—resist being gamed by clever twelve-year-olds.

We are building an AI tutoring agent that can teach any learner anything at any time and at their own pace. Personality, memory, proactivity, and more: the technical constituents of what made Aristotle an incredible tutor thousands of years ago, now built into software.

Mission

Many are racing to build superintelligence. Few are considering the place humans will have in that future. We are.

Join us to work on the most meaningful and urgent challenge of our generation.

Building in Hayes Valley. Raised $5M from leading angels and investors.
By jsv.ai