YuviLab says Israel’s Education Ministry has selected its Spark AI learning platform for English instruction under Project 720, a government-led personalized-learning initiative. Announced earlier in November 2026, the selection adds English to Spark’s existing mathematics and science pilot work, with students able to practice spoken English through conversations adapted to their level.
The selection was described in Deborah Danan’s November 16 JTA article. The English expansion follows government plans to widen AI-assisted language instruction after just 22% of Israeli ninth-graders met national English curriculum requirements last year. Spark’s role is therefore tied to an existing school program and a specific subject, rather than a general-purpose chatbot rollout.
From a family game to school language practice
Founder Moti Malka traced YuviLab to a game he built for his 10-year-old son during wartime school closures. Players had to assemble English sentences correctly to advance in a race. Malka launched the educational platform in March, initially aiming to turn children from consumers of screen content into creators.
The company now offers two distinct products: Spark is paid software for schools, while its student-facing creation platform is free. The latter lets children describe games, apps and learning projects in ordinary language, test what the AI produces and revise their instructions. It supports Hebrew, English, Russian and Arabic.
Student-built projects described by JTA show how that creation platform connects familiar game formats with schoolwork:
- An English escape room required players to solve language questions before they could leave.
- A game modeled on the Iron Dome air-defense system linked successful interceptions to correct arithmetic answers.
- Other children drew on Minecraft and Roblox as templates for learning projects.
Yuvi also displays generated code as children build. Malka said Microsoft education specialists encouraged that feature so students could inspect how their projects work, rather than see only the finished result.
We don’t teach a specific platform
Moti Malka, YuviLab founder, speaking to JTA
Finished work is not the same as understanding
Malka acknowledged that some early users gave vague instructions and accepted polished-looking results even when those results did not match their requests. His stated educational priority is teaching principles: asking better questions, spotting mistakes, criticizing answers and organizing work with AI, rather than mastering a particular tool.
That distinction does not settle whether AI instruction deserves more classroom time. Lea Shahar, a programmer turned high-school coding teacher interviewed by JTA, argued that rapidly changing, intentionally intuitive tools are a poor curriculum priority. She favored more time for interpersonal communication and languages—skills children are less likely to develop alone on a screen.
YuviLab’s school footprint already extends beyond the new English selection. Spark has been used in Project 720’s mathematics and science pilot, and Yuvi works across dozens of local authorities, including Jerusalem, Rishon LeZion and Netanya. Malka says the company’s platforms reach tens of thousands of children and thousands of teachers.
Teachers keep the approval step
On the educator-facing side, Yuvi can create assignments, presentations, tests and interactive learning units. Teachers can distribute the work and identify where individual students struggled; parents have a progress dashboard. The system can then generate a tailored exercise, reducing the need to prepare a different task manually for every child.
Malka describes a restricted-source approach to curbing AI-generated errors. As detailed in the Jewish Times publication of JTA’s article, the teacher-facing platform builds learning units from uploaded Education Ministry textbooks and other approved classroom materials, rather than relying on a general chatbot’s open-ended knowledge.
Teachers can still edit and approve those materials before students receive them. Malka’s stated aim is to remove preparation and administrative work, leaving teachers more time to teach—not to make them accept generated lessons unchanged.
Reader comments
Newest comments first. Replies stay oldest first.