The math problem is simple. A tutor spends two hours preparing materials for a one-hour lesson. That is a ratio of two to one — two hours of invisible labor for every hour of visible teaching. Multiply that by twenty students per week, and you have forty hours of preparation. That is a full-time job on top of a full-time job. The cost is not measured in dollars but in energy, in attention, in the slow erosion of the very enthusiasm that makes a good teacher good. Preply, the world’s largest online language learning marketplace, has calculated that cost. And it has built a tool to eliminate it.
The Threat: Repetition as the Enemy of Connection
The threat is not that AI will replace tutors. The threat is that the repetitive administrative work of teaching will drain the life out of them before they can do what they do best. Preply connects over 100,000 tutors with learners across more than 180 countries, offering one-on-one instruction in more than 90 languages That scale creates a brutal arithmetic. Every tutor, every day, faces the same cycle: write lesson plans, take notes during class, write summaries afterward, create homework, track progress. The human connection that makes language learning work — the motivation, the cultural nuance, the energy of real conversation — gets buried under paperwork. Tutors like Michelle Garcia Ramos know this intimately. Before Preply’s AI feature, she would spend hours and hours prepping for classes and creating homework. That time has now been cut by more than half.
The threat is subtle because it looks like dedication. A tutor who spends hours on preparation is not lazy. They are committed. But commitment has a cost. The energy that goes into writing a lesson summary is energy that cannot go into understanding a student’s frustration with a particular grammar point. The time spent creating homework is time taken away from reflecting on how to explain a difficult concept more clearly. The threat is that the very structure of the job — the ratio of preparation to teaching — slowly grinds down the human element that students say they value most. Preply’s own internal study found that 96% of learners consider human connection and real conversation essential to their progress. The threat is that the administrative burden makes that connection harder to sustain.
The Mechanism: Turning Transcripts into Learning Journeys
The mechanism is elegant in its simplicity. After each one-on-one lesson, OpenAI analyzes the lesson transcript. The analysis is not a generic summary. It is a structured, personalized report that includes key lesson topics, grammar corrections with explanations, vocabulary highlights with translations, pronunciation feedback, and recommended next steps. The feedback is ready within minutes of the lesson ending, appearing in the chat thread where tutor and student can review it together. The mechanism does not replace the tutor’s judgment. It replaces the tutor’s need to manually write all of that from memory.
Preply calls this feature Lesson Insights. It is powered by OpenAI’s API, and it transforms every lesson into a personalized learning journey. The key design choice is timing. Insight generation is scheduled a few minutes before the end of the session, so the feedback is ready for the tutor and student to review together before they say goodbye. This is not a post-hoc report that gets buried in an inbox. It is a live tool that extends the lesson itself. The tutor can look at the AI-generated summary, add their own observations, and discuss the next steps with the student while the material is still fresh. The mechanism turns the end of a lesson from a closing into a launch for the next session.
The mechanism also feeds into Preply’s self-learning exercise engine. The insights from each lesson generate personalized homework. Every human lesson becomes structured practice that compounds over time. Students no longer ask, “How am I doing? Am I getting better? What do I get to do next?” because the system answers those questions automatically. Emily Stott, Staff Product Manager at Preply, explains that students wanted a more tangible view of their growth. Now, with Lesson Insights, the system understands exactly what the student has been talking about, their goals, the topics covered, the tutor’s feedback, and converts that into highly personalized and targeted practice.
The Adaptation: A Cultural Transformation, Not a Tooling Rollout
The adaptation required more than building a feature. It required changing how an entire organization thinks about AI. Preply introduced ChatGPT Enterprise across its organization, running company-wide enablement sessions for more than 600 employees across New York, Kyiv, London, and Barcelona. Weekly active usage grew from 60% to 95%. That is not a coincidence. It is the result of treating AI as a cultural transformation, not a tooling rollout. Leadership set a clear strategy. Teams were trained to use it. AI became part of everyday work across the business.
The adaptation also required picking high-impact use cases. Preply did not try to build fifty AI features. They went deep on one: Lesson Insights. The logic is straightforward. It is better to have one feature with clear user value and measurable impact than many experiments that never reach quality. The results speak for themselves. 75% of English-language learners actively use Lesson Insights. More than 70% of tutors use the feature. Around 75% of active learners continue engaging with Lesson Insights more than a year after adoption. The satisfaction rating is 4.7 out of 5, based on more than 300,000 ratings received directly on the platform. The product market fit score is 70%, significantly above the threshold often associated with strong customer demand.
The adaptation also required building partnerships. Preply evaluated multiple AI models across language-learning tasks before settling on OpenAI. Dmytro Voloshyn, Co-founder and CTO, explains that OpenAI provides state-of-the-art models that solve problems for customers. The partnership is now at the center of Preply’s ecosystem and how the company operates. OpenAI’s API is used across the organization, supporting customer service workflows, connecting internal systems, and helping teams automate operational processes. Custom GPTs, like the Brand Voice GPT, are embedded into daily workflows, helping teams create content that stays consistent with Preply’s tone, messaging, and brand standards.
The adaptation also extended to engineering. Around 94% of Preply engineers use Codex and AI coding assistants for code generation, PR reviews, debugging, and accelerating development workflows. Dmytro Voloshyn says Codex helps engineers write code with a speed that was previously unseen. With such a powerful tool, engineers can focus more on architecture in the system as a whole and on solving customers’ problems. By reducing time spent on routine coding tasks, engineering teams can focus on building new learning experiences, scaling products across languages, and delivering value to tutors and learners faster.
The adaptation was not just internal. Preply hosted its fourth global online teaching conference to power tutor progress and shape the future of language learning. The conference brought together tutors from around the world to share best practices, learn about new tools, and discuss how AI is changing their work. The message was clear: AI is not a threat to tutors. It is a tool that makes them better at what they do. The conference was not about selling a product. It was about building a community of practice around AI-enhanced teaching.
The Hope: More Time for What Matters
The hope is that AI does not make tutors obsolete. It makes them more human. By eliminating the repetitive administrative tasks, AI frees up time and energy for what really matters: motivation, engagement, energy, cultural nuance, and real conversation. Preply’s co-founder and CEO Kirill Bigai puts it simply. There is an amazing opportunity to combine what humans are the best at — motivation, engagement, energy — with AI as a copilot. AI can handle all of the repetitive and administrative tasks.
The hope is visible in the numbers. Preply has tripled its revenue in the last five years. It has tripled the number of tutors on its books since its previous funding round three years ago. It has expanded its offerings with more than 40 new languages. In January 2026, Preply raised $150 million in a Series D round led by WestCap, bringing its total funding to $299 million and its valuation to $1.2 billion. The global direct-to-consumer language learning market is forecast to reach $227 billion by the end of 2025, with about 1.8 billion people actively trying to learn a new language.
The hope is also visible in the stories of individual tutors. Michelle Garcia Ramos says that before she started using Preply’s AI feature, she would spend hours and hours prepping for classes and creating homework. Now that time has been cut by more than half. The benefit is not simply efficiency. AI provides ideas, highlights learning opportunities, and helps identify what students should focus on next. The result is better preparation, more focused lessons, and faster progress for learners. The hope is that every tutor can have that experience — that the ratio of preparation to teaching can shift from two-to-one to one-to-one, or even better.
The hope extends beyond individual tutors to the entire learning experience. Preply reports that most active learners still engage with Lesson Insights more than a year later. Both tutors and students provide strong product feedback, boosting overall retention. Emily Stott acknowledges that AI-powered tools might have an element of novelty. But if people are coming back and engaging months later, time and time again, that is a very strong signal of value. Learners and tutors find value in the solutions.
The hope is also about scale. Preply wants to expand its AI and data capabilities and grow its product and engineering teams so it can reach more learners and tutors globally. WestCap, which previously helped to scale up global marketplaces such as Airbnb and StubHub, will provide invaluable expertise. The goal is to reshape the way people learn second languages. Preply co-founder and CEO Kirill Bigai feels a deep responsibility regarding the company’s role in shaping global learning experiences. He believes the company is doing a very good job of it. They connect people with the world’s best tutors, amplified by AI, bringing learning efficiency to a level that was previously unreachable.
The hope is not naive. Preply is clear that AI is not going to turn to automation at the expense of humans. The focus remains on using AI simply to enhance tutor-driven learning and support. The internal study from last year found that 96% of learners consider human connection and the ability to engage in real conversations with another person as essential to their progress.
The Deepest Question
A child might ask: If the AI does all the preparation and the homework and the feedback, what does the tutor actually do? The question is not naive. It cuts to the heart of what makes language learning work. The answer is not in the technology. It is in the relationship. The tutor provides the energy that makes a student want to keep going when the grammar is hard. The tutor provides the cultural context that makes a phrase come alive. The tutor provides the motivation that turns a lesson into a habit. The tutor provides the human connection that makes learning a language about more than words.
The deepest question is not about what AI can do. It is about what humans can do when they are freed from the repetitive tasks that drain their energy. The answer is not yet fully known. But the early signs are promising. Preply’s tutors are spending less time on paperwork and more time on teaching. Their students are staying engaged longer and making faster progress. The ratio of preparation to teaching is shifting.
The math problem at the beginning had a simple answer. Two hours of preparation for one hour of teaching is unsustainable. It burns out good teachers. It reduces the quality of instruction. It makes the human connection harder to sustain. The solution is not to eliminate the preparation. It is to automate it. Preply has shown that it is possible. The question now is whether other industries will follow. The deepest question is whether we will choose to use AI to make humans more human, or whether we will use it to make them obsolete. The answer is not in the technology. It is in the choices we make.
Sources
1. Preply
2. WestCap
3. Airbnb
4. StubHub