Sparkli Raises $5M for AI Learning for Kids

Sparkli, a Zurich-based generative AI startup, has raised $5 million in pre-seed funding led by Founderful to build immersive, interactive learning experiences for children. Founded in 2025 by former Google colleagues Lax Poojary, Lucie Marchand, and Mynseok Kang (also reported as Mynseok Kang in several reports), Sparkli is positioning its product as a multimodal alternative to text-heavy AI chatbots for kids aged five to twelve.

$5M pre-seed and the team behind Sparkli

The $5 million pre-seed round was led by Founderful, with Sparkli highlighting the financing as a key step toward scaling its multimodal learning engine for children. In coverage of the deal, Founderful’s founding partner Lukas Weder pointed to the founders’ technical depth and the broader need for more engaging, real-world learning experiences for kids as the rationale behind the investment.

Sparkli was founded in 2025 by Lax Poojary, Lucie Marchand, and Myn Kang, and the team is frequently described as “ex-Google” because of their prior work history. Multiple reports identify the company as Zurich-based and describe its focus as building a “multimodal AI-native learning engine” designed specifically for children aged 5–12.

That “built for kids” positioning matters, because many general-purpose AI tools still default to long text outputs that are hard for younger learners to process without adult help. Sparkli’s core bet is that children learn best when they can explore, make choices, and stay inside an interactive experience rather than being pushed into passive reading or one-way explanations.

For education stakeholders, this round is also a signal: early-stage capital is continuing to move into applied generative AI, not just for productivity at work, but for structured learning outcomes in classrooms and at home. As the ai world organisation tracks this shift across the ecosystem, stories like Sparkli’s provide a concrete case study for how “genAI + education” is evolving from chat interfaces into richer, multimodal learning products that could show up in future showcases at the ai world summit and related ai world organisation events.

What Sparkli is building: “learning expeditions” in real time

Sparkli’s product is described as generating interactive, multimedia learning “expeditions” tailored for children aged five to twelve. Instead of only responding with text or a single audio narration, the platform creates a mix of formats in real time—audio, video, images, quizzes, and games—so kids can learn by interacting, not just consuming.

In practical terms, this approach aims to turn a child’s curiosity into an evolving learning path that feels closer to discovery than to traditional instruction. Reports about the product describe it as a multimodal experience built to keep engagement high through visuals, voice, gamification, and simulations, which is a deliberate contrast to text-only or prompt-and-answer learning.

The “expedition” framing also implies a more narrative, journey-like learning flow: children start with a question or topic, and the system builds a connected set of activities around it rather than giving a single answer and ending the session. That structure matters in K–12 learning because attention is not just a constraint—it’s often the difference between a child retaining a concept and dropping it after a few seconds.

A notable part of Sparkli’s positioning is that it aims to transform “screen time” into a more purposeful, active learning experience. In a market where parents and schools increasingly worry about passive consumption, the company is leaning into the idea that devices can be used for guided exploration—especially when content is generated and adapted in the moment to match how children engage.

This shift toward multimodal learning is also a broader industry trend worth tracking for ai conferences by ai world, because multimodality changes what “personalization” means. It’s no longer only about changing the reading level of a paragraph; it can mean changing the medium entirely—turning a concept into an image-based explanation, a short interactive simulation, or a game-like quiz loop that reinforces understanding.

From the perspective of the ai world organisation, this is exactly the kind of product category that deserves debate and due diligence at the ai world summit, including questions around age-appropriate UX, content safety, learning measurement, and the role of teachers as facilitators. As we prepare programming themes for ai world summit 2025 / 2026 and other ai world organisation events, Sparkli’s model is a strong example of where generative AI is heading: interactive, child-centric, and designed to produce learning experiences—not just answers.

Pilots, partnerships, and the 2026 rollout plan

Sparkli is not describing its first phase as purely direct-to-consumer; instead, it has been piloting through school networks at meaningful scale. The company is currently validating its platform through a strategic pilot with one of the world’s largest private school groups, and reporting says this partner network includes more than 100 schools and over 100,000 students.

This kind of institutional pilot matters because it forces an edtech product to work in real classroom conditions, not only in controlled demos. At school scale, questions change: teachers need clarity on how the tool fits a lesson, administrators need to understand oversight and policy alignment, and parents need confidence that the content is appropriate and useful.

In terms of timeline, Sparkli has communicated that it plans to expand globally through educational partnerships before opening broader consumer access in 2026. Separately, company announcements and reporting have also referenced preparation for a private beta launch in January 2026 as it scales the generative learning engine.

The broader rollout strategy is worth noting: Sparkli appears to be using schools as a proving ground, then moving toward families once product-market fit and classroom usability are stronger. This sequencing can help a young company harden its safety systems and learning design because school environments tend to demand repeatable structure and clearer guardrails than purely at-home use.

For partners, the “education first, consumer next” plan is also a signal about how Sparkli may commercialize: likely a mix of institutional subscriptions and parent-facing plans, a path that has precedent in K–12 software markets. If that strategy works, it could create a feedback flywheel where classroom usage improves the product, and family usage expands the data on how children learn across different contexts.

For the ai world organisation, these partnership-led go-to-market strategies are valuable content themes for the ai world summit and ai conferences by ai world, because they sit at the intersection of AI product design, education policy, and stakeholder trust. As ai world summit 2025 / 2026 conversations increasingly focus on responsible deployment—not just capability—Sparkli’s approach offers a real example of how startups can validate learning technology with institutions before scaling to mass consumer audiences.

Why this funding matters for AI + education (and what leaders will debate)

Sparkli’s $5M pre-seed round is not simply a headline about another startup raising money; it’s part of a bigger shift in how generative AI is being packaged for learning. Early genAI education products often resembled “chatbots for homework,” but Sparkli is pushing toward immersive experiences that blend multiple modalities in real time—audio, video, visuals, and interactive checks for understanding.

That design direction matters because it changes what success can look like. A chatbot can answer a question correctly and still fail as a learning tool if the child disengages, misunderstands, or cannot connect the idea to the real world. By contrast, a well-designed interactive expedition can scaffold curiosity: it can show, ask, challenge, and adapt—ideally keeping the child active inside the learning loop rather than merely reading the output.

At the same time, multimodal genAI for children raises high-stakes questions that the market must address responsibly. If a system generates images, video-like explanations, quizzes, and games, then content safety and age-appropriateness become more complex than filtering a text response. Classroom deployment also raises governance questions: teacher controls, auditability, and alignment with curriculum goals. These are the kinds of issues that belong on stage at the ai world summit, especially as organizations, regulators, and educators look for best practices that scale.

There is also the question of measurement: “engagement” is not the same as “learning.” A product can be captivating and still produce shallow understanding, so schools and parents will look for signals that knowledge is retained and transferable—such as whether children can explain concepts in their own words, apply ideas to new problems, or build projects that demonstrate understanding. Early pilots described in reporting suggest Sparkli is being tested in classroom-like contexts, which can help generate more meaningful evidence over time.

Another reason the round stands out is the founder profile and the investor thesis. Founderful’s comments, as reported, emphasize both the team’s technical capability and the demand for more engaging learning experiences—suggesting the bet is as much about product execution and UX for kids as it is about model performance. In a crowded generative AI market, “how it feels to use” often becomes a major differentiator, particularly for children, where motivation and ease-of-use are core to whether learning even begins.

From the ai world organisation standpoint, Sparkli is also a useful “innovation watch” item for ai world organisation events because it sits at a crossroads: education, consumer product design, multimodal AI, and childhood development. For ai world summit 2025 / 2026 editorial programming, this category invites strong panel formats—startup founders paired with school leaders, child psychologists, policy experts, and safety researchers—to pressure-test what “interactive genAI learning” should look like when deployed at scale.

Finally, Sparkli’s roadmap—pilots through school networks, partnership expansion, and consumer access in 2026—reflects a practical understanding of trust-building. In education, trust often arrives through demonstrated reliability: teachers seeing outcomes, parents seeing meaningful curiosity, and institutions seeing guardrails and governance. If Sparkli can prove these elements while scaling, it may help define a model for how the next generation of child-focused AI learning platforms are built.