Yandex Turns Its Search Engine Into a Tutor With a New AI Learning Mode
Yandex has launched an experiment that redefines how homework works: a new “Try Learning Mode” button now appears directly on math problem search pages.
A student struggling with an equation can now open a search engine, type in the problem and meet something very different from a standard answer. Instead of delivering a finished solution, the system starts a dialogue: “Let’s think this through together. What do we already know? How can we apply this formula?” That is the experience Yandex is currently testing.
Formula for Understanding
Inside the search interface, within the Alisa assistant, users now see a “Try Learning Mode” button. This is an unusual role for AI. Instead of acting as an all-powerful assistant that does everything for the user, Yandex is positioning it as a mentor that guides the learning process. The interface includes a workspace for drafts and a built-in calculator, designed to support step-by-step problem solving.
The feature is currently available in a limited test rollout, the company said. It is expected to appear soon in chats with Alisa AI, powered by an updated large language model developed by Yandex.
Beyond School Use
At first glance, this may look like a routine feature update. But it signals a broader shift. Russian big tech companies are moving beyond generic chatbot functionality and embedding neural networks into everyday, high-impact services.
The “smart tutor” is expected to reach millions of users. Darya Kozlova, Director of Education at Yandex, says the system is designed specifically as a tutor for students and their parents, helping them prepare for exams and build skills.
The technology also reflects real demand. Educators are already integrating AI into their workflows. According to a survey of university instructors, 66% use AI in their work. About half rely on it to generate images, create tests and assignments, and prepare presentations.
Do No Harm, Teach Instead
In 2024, Yandex Uchebnik (Yandex educational platform) introduced an AI-powered math assistant for grades 5 through 8. It guided students through problems using prompting questions. At the time, the tool operated within a dedicated learning platform. Now, the same “do no harm, teach instead” approach is moving into core user environments such as search and Alisa.
In the same year, the company began training students and educators to work with neural networks. By 2026, a school course titled Vvedeniye v iskusstvennyy intellekt (Introduction to Artificial Intelligence) was added to the federal curriculum, making AI part of formal education.
One of the most significant developments, however, took place outside the digital environment. Minister of Education Sergey Kravtsov, Minister of Digital Development Maksut Shadayev and ASI head Svetlana Chupsheva met with market leaders including Yandex, Sber and others. At VEB.RF laboratories, they began what participants described as a “historic conversation.” Kravtsov framed the discussion as the starting point for defining what role AI should play in schools and how to ensure it does not replace knowledge. Shadayev called for structured implementation and clear criteria. Chupsheva summarized the challenge: “The key issue today is mitigating the risks associated with AI technologies. It is critically important – as the president has emphasized – that AI adoption does not lower the quality of education, that children do not lose the ability to think and analyze information or rely on easy answers. Our shared task is to ensure that technology improves the effectiveness of the educational process.”
A Code for Digital Mentors
Embedding an educational scenario into Alisa shows how large language models are evolving from abstract tools into technologies with clear, practical value. Homework is not a niche issue but a universal one, and Russian developers are keeping pace with – and in some areas moving ahead of – global trends by building fully integrated EdTech assistants rather than isolated features.
Parents no longer need to revisit complex math concepts or hire tutors, while students can access guided support at any time. Humanities students at HSE University have already used YandexGPT to write coursework, relying on it to find information, structure content and correct errors. Around 43% of university students already use AI in their studies, and that number is expected to grow.
At the same time, governance matters. Institutions need clear criteria to distinguish a helpful digital mentor from a tool that enables cheating. Universities are already introducing ethical guidelines for AI use. For example, ITMO University is developing competency frameworks and digital assistant services to reduce the workload on faculty.
What comes next? The underlying logic of the service is universal. Similar assistants for physics, chemistry or language learning are likely to follow. Math is just the starting point.
