Sparkli Raises $5M for Interactive AI Learning for Kids

Sparkli raised $5M pre-seed led by Founderful to build multimodal AI learning expeditions for ages 5–12, piloting across 100,000+ students worldwide in 2026.

Sparkli Secures $5M Pre-Seed to Build Interactive Generative AI Learning for Kids

Sparkli has raised $5 million in pre-seed funding, led by Founderful, to build an interactive, multimodal generative AI learning platform for children aged 5–12.
From the Global Martech Alliance viewpoint—where the focus is helping teams discover, evaluate, and adopt the right tools—this round is worth watching because it points to a fast-emerging category: AI-native learning experiences designed around engagement, personalization, and safe-by-design product decisions.​

Funding snapshot: who backed Sparkli and why it matters

Sparkli was founded in 2025 by former Google employees Lax Poojary, Lucie Marchand, and Myn Kang (also referenced as Mynseok Kang in some coverage).
Multiple reports describe Sparkli as Zurich-based and emerging from stealth with this $5M round.
The round is widely reported as being led by Founderful, with participation that includes Arc Investors, and Sparkli also received a grant from Innosuisse.​

Investor rationale has been consistent across coverage: the founding team’s technical depth, plus the belief that children need learning that feels more “real” and active than static content or text-only chatbot experiences.
In comments attributed to Founderful partner Lukas Weder, the emphasis is on both engineering capability and pedagogy—plus early traction with school pilots as proof that the product direction matches a real classroom need.​
From an ecosystem lens, that traction is the signal: distribution in education rarely starts with consumer growth loops; it starts with institutional credibility, pilot outcomes, and partnerships that can survive long procurement cycles.

What makes this round particularly interesting is the timing: Sparkli is positioning itself not as “another AI tutor,” but as a more immersive, multimodal interface that’s meant to meet children where learning already happens—through visuals, play, simulation, and guided exploration.​
That design choice matters because it reframes generative AI from a “Q&A machine” into an “experience engine,” which is a category shift we’re also seeing across martech—moving from content generation to content orchestration across formats and touchpoints.​

Inside Sparkli: “expeditions” instead of lessons

Sparkli’s core product idea is the learning “expedition”—interactive sessions that turn a child’s question into a multimedia journey rather than a text response.
Coverage describes the platform generating audio, video, images, quizzes, and games in real time, enabling children to explore topics through interaction instead of relying only on reading or voice prompts.​
Sparkli has also positioned itself explicitly as an alternative to text-based AI chatbots by leaning into visuals, voice, gamification, and simulations.​

Several examples reported in the press show how Sparkli wants kids to explore “big curiosity” questions (for instance, Mars scenarios) by interacting with the concept rather than consuming it passively.
In at least one description of the product experience, content is broken into chapters that combine narration, images, videos, quizzes, and simple games—while “choose-as-you-go” adventures encourage exploration without the pressure of being graded.​
This “remove performance pressure, increase curiosity” framing is central to Sparkli’s positioning and aligns with the broader edtech push toward mastery, agency, and intrinsic motivation rather than rote correctness.

Some reporting also suggests Sparkli is building personalization through an evolving interest/knowledge graph for each child, so future expeditions can adapt to what the learner has explored and enjoyed.​
If that approach holds up in practice, it’s not just personalization in the “recommendation” sense—it’s personalization in the “learning path design” sense, where sequencing, difficulty, and format can change dynamically.​
For product teams, this implies a bigger technical lift than typical “AI wrapper” apps: multimodal generation, guardrails, evaluation, and telemetry all need to work together with near-zero friction.

From a Global Martech Alliance angle, the key takeaway is that “multimodal” isn’t a buzzword here—it’s the entire UX strategy.​
When the user is a child, attention and comprehension don’t behave like adult productivity workflows, so Sparkli’s bet is that interactive media and guided play can make learning feel closer to discovery than instruction.
That’s also why this round deserves attention beyond edtech: it’s a blueprint for how generative AI products may need to be built when the end user’s expectations are shaped by games, short-form video, and interactive storytelling.

Go-to-market: schools first, consumers later

Sparkli is currently piloting with school networks reaching more than 100,000 students, and at least one report describes a strategic pilot with one of the world’s largest private school groups spanning 100+ schools and 100,000+ students.
This school-first approach is typical for products that need controlled environments, adult oversight, and proof of learning outcomes—especially when the user is a minor.
It also creates a practical testing bed for safety, classroom usability, and teacher workflows before the product faces the unpredictability of open consumer usage.​

On timing, Sparkli has been reported as preparing for a private beta launch around January 2026, with broader consumer access expected later (often cited as 2026).
That sequencing (institutional validation → beta → consumer) is a deliberate risk-management move, because the most painful failures in kids’ products aren’t purely technical—they’re trust failures that quickly become reputational and regulatory problems.

Some reporting adds that teachers are already using Sparkli in multiple ways during pilots: quick exploration at the start of class, discussion-led learning formats, and even generating follow-up work so students can deepen understanding after a concept is introduced.​
That detail matters because it hints at where Sparkli could become “sticky” inside schools: not replacing educators, but accelerating lesson setup and helping create differentiated learning moments.​
In martech language, this is workflow adoption: a tool wins when it fits the operator’s day, not when it only looks impressive in demos.​

Globally, Sparkli’s expansion is described as happening via educational partnerships before consumer access opens more broadly.​
That points to a partnership-heavy growth model—school groups, curriculum-aligned collaborators, distribution partners, and potentially device ecosystems—rather than pure app-store marketing.
For teams evaluating edtech tools, it also means vendor due diligence will likely focus on deployment readiness (training, controls, reporting) as much as on the child-facing experience.

What this signals for MarTech, Data, and CX teams

Global Martech Alliance’s mission emphasizes helping teams choose tools with clarity—through comparisons, reviews, and practical resources—because modern stacks are crowded and evaluation cycles are expensive.​
In parallel, the MarTech, Data & CX Council on the same domain describes a landscape “cluttered with 8,000+ solutions,” reinforcing the reality that tool sprawl makes confident decisions harder.​
Sparkli’s rise fits that broader theme: we’re watching generative AI create entirely new tool categories faster than governance models can keep up.

Even though Sparkli is education-focused, the product philosophy maps directly to where martech is going:

• Experience-first generation, where content is produced as an interactive flow (not just assets).
• Format orchestration, where audio, visuals, and interactivity are generated and sequenced dynamically.
• Personalization with memory, where the system adapts based on user interaction over time (often described via interest/knowledge graphs).​

For CX leaders, Sparkli is also a case study in a bigger shift: “engagement” isn’t only about better messages—it’s about building interfaces that invite participation.
This is especially relevant as brands experiment with conversational commerce, AI shopping assistants, and interactive product discovery; the winners won’t be the ones that can generate the most words, but the ones that can guide the user through a satisfying journey.​
In that sense, Sparkli’s “expedition” framing is a helpful metaphor for martech: reduce friction, increase agency, and keep the user moving forward with clarity.

There’s also a strategic data lesson here. If Sparkli is indeed building evolving profiles of interests and knowledge, that resembles what martech teams think of as “progressive profiling”—but in a context where the ethical bar is far higher because the user is a child.​
For marketers and product builders, this is the direction of travel: personalization will increasingly depend on long-lived user models, but trust will depend on what data you collect, how transparently you explain it, and how tightly you control it.
Kids’ AI products simply force those questions earlier—and more publicly—than adult-focused tools.

Finally, Sparkli underlines a market truth: generative AI is moving from “content generation” to “learning and decision support,” where the output isn’t a post or an image—it’s a changed understanding in the user’s mind.
That’s a higher-stakes promise than most marketing automation platforms make, and it will require stronger evaluation methods (quality, bias, safety, consistency), not just “does it work.”
For tool buyers, the implication is clear: RFPs and vendor assessments will need to evolve to measure “experience outcomes,” not only feature checklists.

The non-negotiables: safety, privacy, and child-first AI governance

Any product directed to children must treat privacy and consent as core product requirements—not as legal footnotes.
In the U.S., COPPA imposes requirements on operators of online services directed to children under 13, including obligations tied to collecting personal information from that audience.​
Summaries of COPPA’s intent and requirements emphasize parental notice and verifiable parental consent before collecting, using, or disclosing children’s personal data.

For AI-native learning tools, the governance surface area expands because the system isn’t only collecting data—it’s generating content dynamically.
That means safety has multiple layers: content policy (what’s allowed), model behavior (what it tends to produce), UX constraints (what the child can do), and reporting/controls for adults.
It also means schools and parents will ask harder questions about data retention, third-party processors, training-data usage, and whether children’s interactions can be used to improve models.

In practice, the most durable AI learning platforms will likely be the ones that operationalize “trust” as a feature set:

• Clear parent/administrator controls and visibility into what’s being generated and consumed.
• Strong content guardrails and age-appropriate prompting pathways, especially for open-ended exploration.
• Measurable classroom utility (teacher-friendly setup, reliable outputs, and predictable behavior during lessons).​

Sparkli’s pilot-led approach gives it a potentially valuable advantage here: schools can stress-test the product in real environments where adults can observe what happens, not just what’s promised.
And because Sparkli is emphasizing “agency” and curiosity, it will need to show it can deliver that freedom without slipping into unsafe or unhelpful content pathways—an especially difficult balance for generative systems.
The startups that solve this balance won’t just win in education; they’ll set expectations for how interactive AI products should behave across industries.