We are a small London startup with the ambition to change education with ML-powered tutoring. Our flagship product is a mobile application for teaching English to intermediate and advanced learners.

We’re on the verge of solving one of the biggest challenges in education – making high-quality, personalized learning accessible to everyone. What was once technologically impossible is now within our grasp.

Why?

In a world where ML & AI are advancing at maximum speed, the question on everyone’s mind is: How do we as human minds stay relevant?

The answer lies in education.

However, we face a challenge: traditional education models have added more years of study as the body of knowledge has grown. 150 years ago, three years of study was standard; now it’s 15 years. We can’t keep stretching this timeline indefinitely.

The solution: Smarter, more efficient learning methods that keep pace with our rapidly evolving world.

OK, but how?

1. Effective methods actually exist and have always been around.
2. One-on-one tutoring, like the Socratic method or the Oxford/Cambridge model, has proven to be incredibly effective.
3. In fact, tutoring is at least twice as effective as the traditional classroom setup.
4. The catch? Tutoring is expensive and hard to scale – there aren’t enough good tutors to go around.
5. Tutoring method in a nutshell: tutors assess student's current knowledge and select next optimal activity.
6. Modern AI methods can help to replicate and scale this tutoring approach.
7. Think about tutoring in Reinforcement Learning terms:
   • the tutor (RL: agent) teaches student (RL: environment);
   • it checks understanding and motivation (RL: state);
   • then it selects the next learning activity (RL: actions);
   • based on teaching strategy (RL: policy).
8. Motivation is an emergent property of the system, RL will automatically extract the importance of motivation for long-term learning outcomes.
9. And now, we finally have the technology to make this a reality: modern LLMs, GenAI, voice-to-voice models. It was impossible a few years ago, it is possible now.
10. The goal: create an AI tutor that’s as effective as the best human tutors out there.
11. This system will learn and improve from billions of interactions, constantly experimenting and evolving.
12. The ultimate goal: superhuman tutoring ability, leveraging millions of student journeys.