For investors

A multimodal learning system that has cleared substantive examination

The system merges what is normally handled separately — video, speech with its intonation, and the course of the dialogue — into a single model of the learner, and changes how it teaches from that model rather than from a test at the end of a topic.

Stage: there is no product yet.

A first version and a pilot with 20–30 students are in preparation. Of the fourteen features in the patent claim, none is fully implemented: three are in development, three partial. We say this first so as not to waste your time.

Legal position

On invention application No. 2025118773, «Intelligent system of personalised learning with multimodal analysis and adaptive interaction», the Russian patent office (Rospatent) has issued a decision to grant a patent. The solution passed substantive examination: novelty, inventive step, industrial applicability.

Filing date and priority
07 July 2025
Claim structure
1 independent + 14 dependent
International application
PCT/RU2026/050211

To be precise: the patent has not been granted yet — state registration is under way, so no patent number exists and we cannot yet speak of a granted patent. Patent pending (RU). A Russian patent has effect in Russia only; the PCT application opens national phases abroad, but there is no such thing as an «international patent».

What the protection covers

The independent claim, stated plainly. Protection attaches to these features, not to the product as a whole.

  • A multimodal coordination core: vision, speech and dialogue converge into one model of the learner instead of being processed separately
  • Speech analysis with prosody — intonation and pace, not just recognised text
  • Adaptation for autism, ADHD and dyslexia with switching between presentation modalities
  • A dialogue spiral: returning to what was not absorbed, with memory management across sessions
  • Protégé mode: the learner explains the material back to the system, inverting the roles
  • Autonomous generation of interface elements driven by the learner’s current state

Market

Russian — because that is where protection will have effect, where the pilot runs and what language the product speaks.

₽154 bn
Revenue of Russia’s top-100 EdTech, 2025
+12%
Growth during 2025
32 → 19 → 12
Growth rate, % in 2023, 2024, 2025

We show the third figure deliberately: the market has been decelerating for three years, and betting that it will carry the product on its own would be dishonest. Our bet is not on market growth but on the audience it serves worst. Source: Smart Ranking, 2025 results.

What the pilot tests

The pilot tests a hypothesis; it does not ship a product. What exactly it tests was written down before the work started.

Hypothesis

Whether the loop «observe the state → adapt the delivery → return to what was missed → have the learner explain it back» works on real people and delivers perceived value.

Sample and material

20–30 students, school algebra: four linked topics with the links made explicit, because the spiral works by returning to earlier material.

How it runs

A single process, browser client, laptop with camera and microphone. Processing happens on the participant’s device; streams are not retained after the session.

What counts as a result

Qualitative feedback from participants. A sample this size cannot yield a statistically significant effect, so we neither promise nor measure double-digit reductions in learning time.

Sign language recognition is not part of the first version: it depends on sign language datasets and runs as a separate track with its own timeline.

Team

Three people. The first two are the inventors.

Alexey Kulikov

Founder, lead developer

Author of the idea and of the AX-architecture; responsible for engineering.

Anastasia Kislitsyna

Co-founder, education expert

Lecturer at SPbSUITD. Educational technology and pedagogy. Co-inventor.

Alexander Tsarev

CEO

Representative on the invention application; runs the legal track of the project.

Let’s talk specifics

There is no form on this site: we do not collect personal data through it. Write to us — briefly, who you are and what interests you. Round terms and timelines we discuss by email rather than publish: with no product yet, any public figure would be invented.

info@synapster.ru