Before ChatGPT: How a Lithuanian EdTech Startup Dared to Reinvent Education with AI
- Krzysztof Kosman
- 13 hours ago
- 3 min read
In late 2022, the world was still waking up to the implications of large language models. ChatGPT hadn't yet become a household name. Policymakers were not scrambling to regulate AI in schools. And yet, in Lithuania, two unlikely co-founders — a seasoned education consultant and a mathematically inclined AI specialist — were quietly sketching out the blueprint of what would become EditAI, a tool designed to help teachers craft engaging, individualized lessons at scale.
Their story is not one of being in the right place at the right time. It’s a story of being too early — and choosing to build anyway.
🎧 Listen to the full story on EdTech Dots Podcast
From Car Rides to Slideware
Kristina Jonkuvienė and Kęstutis Jovaišas didn’t set out to start a company. They set out to solve a problem. After visiting over 400 schools across Europe as consultants, they saw firsthand how rapidly education was falling behind the pace of technological change.
Teachers were overwhelmed. Curricula were rigid. Classrooms were fractured between students who needed different paces, different languages, and different lenses on the same subject matter. And yet the tools available to teachers hadn’t evolved.
So they began an unusual process of ideation: every school visit became an experiment, and every long car ride between schools was a time for synthesis. The idea took shape not in a pitch deck or a hackathon, but in conversation, reflection, and repeated exposure to pain points. They drafted their first prototype on slides, long before there was any code.
This was before ChatGPT.
Manual AI: Prototyping the Impossible
How do you build an AI product when the AI doesn’t yet exist? You do it manually.
Kristina and Kęstutis began creating sample lesson materials — manually generated, carefully crafted outputs meant to simulate what an ideal AI assistant might produce one day. Then they took those mock outputs to teachers and asked a simple question: Would you use this?
The answer: overwhelmingly yes. Teachers wanted these lesson scaffolds. But they also needed thousands of them. And the manual approach? It took over a month to create just one.
Then ChatGPT launched.
The Inflection Point
At first, the team was skeptical. Could ChatGPT — then shaky in its support for smaller languages like Lithuanian — replicate the quality of their handcrafted lessons? The answer, after weeks of testing and prompt iteration, turned out to be yes. Eventually.
In the early days, AI-generated content scored just 2.5/10 in teacher assessments. But by treating each batch of feedback as a learning opportunity — not just for the model, but for the prompt engineers and product team — they reached an average score of 9.5. This wasn't a product built on LLMs; it was a product tuned with teachers, from the ground up.
And that made all the difference.
Bravery, not Brilliance
What stands out about EditAI’s origin is not some eureka moment of genius, but a series of persistent acts of courage. Courage to build something with no guarantee of technical feasibility. Courage to return to angry teachers on a national holiday just to gather feedback. Courage to believe that imperfect tools would improve — and that it was worth building the workflows ahead of the algorithms.
In many ways, EditAI is not an AI product, but a belief system: that education can be made more human by leveraging machines — not to replace teachers, but to return their time and attention to students.
Lessons for an EdTech Startup
There’s a pattern emerging in the best AI-native EdTech startups: they rarely start with code. They start with fieldwork, with deep listening, and with radical proximity to the classroom.
EditAI reminds us that technological inflection points don’t just reward the technically brilliant. They reward those who are willing to build bridges between what’s needed and what’s possible — even before those bridges are structurally sound.
What if more founders worked this way?What if more products started with school visits instead of slide decks?
EditAI didn’t wait for ChatGPT to validate its mission. It anticipated it. And in doing so, it demonstrates a rare kind of foresight: not just into what AI can do, but into what education needs. As systems struggle to evolve, maybe the most important innovation is not the technology itself — but the persistence of those who dared to build before it was obvious.
