SpeEdLabs Survives the Edtech Winter to Grow 6x

Hand the class topper two easy questions and they learn nothing. Hand an average student two impossible ones and they learn nothing either, except perhaps that they're bad at physics. Vivek Varshney, founder of SpeEdLabs, built an entire company on that deceptively simple observation. Then he had to keep it alive through the worst funding drought Indian edtech has ever seen.

Today, around 1.5 lakh students across roughly 700 schools use SpeEdLabs' books, apps, assessments and teacher tools. More than 80% of those schools are in small towns. Its newer brand, STEMLearn.AI, runs robotics, coding and AI programmes and a national AI Olympiad. And after a boom-era stumble that Vivek is refreshingly candid about, the company has grown six times in two years and is targeting profitability in FY27.

From Kasganj to Kota to Private Equity

Vivek grew up in Kasganj, a small town in western Uttar Pradesh, and did his entire schooling there. That detail matters more than it seems, because the company he'd eventually build would find most of its customers in towns that look a lot like it.

From Kasganj he went to IIT Kanpur for mechanical engineering. Then, before most of his batchmates had settled on a direction, he spent time as a teacher at Career Point in Kota — the city that has become shorthand for India's competitive exam machine. After that came an MBA from IIM Lucknow and roughly 11 years in finance, across equity research at Morgan Stanley and private equity and structured debt at Deutsche Bank and ICICI Venture.

It's an unusual combination. Plenty of edtech founders have taught. Plenty have done finance. Fewer have done both, and that pairing shows up everywhere in how he talks — the teacher's instinct about what actually helps a student, and the investor's instinct about what actually keeps a company solvent.

The Proximal Zone

The core thesis of SpeEdLabs is that every student deserves a chance to improve from their own level, not from some imaginary class average.

Educators will recognise this as a practical version of Vygotsky's "zone of proximal development" — the band of difficulty where a learner is stretched but not broken. Vivek calls it the proximal zone of learning, and the whole product is engineered to keep each student inside it.

What's interesting is how he describes the engine. He doesn't reach for "AI" as the headline. He says most of the personalisation is driven by algorithms and classical machine learning, with generative AI mainly used on the content side. In a market where every pitch deck has become an AI pitch deck, that's an honest distinction.

The payoff Vivek cares about isn't a single test score. It's habit. Easy questions waste a strong student's time. Brutal questions demoralise a struggling one. Questions pitched just right give students enough wins to keep practising, and practice is the thing that actually moves outcomes.

Proprietary Logic, Off-the-Shelf Plumbing

Asked whether the models are built in-house, Vivek draws a clear line between two layers. The logic — the maths of how questions are sequenced for each student and why — belongs to SpeEdLabs. It came out of a core team of more than 25 to 30 senior teachers and professionals, many from IIT backgrounds, who encoded their own classroom experience into the rules.

The technical plumbing, on the other hand, uses whatever the best existing models and tools are. There's no ego in rebuilding what already works.

Built for the Aspiration Economy

SpeEdLabs started where Indian student anxiety is most concentrated: competitive exams. Almost everyone wants a good college and a good placement, and those aspirations funnel students after Class 12 into a gauntlet of entrance tests — JEE for engineering, NEET for medicine, IPMAT for management, law entrances, and a long tail of national and private university exams.

The distribution model is what makes it work at scale. Many schools now run preparatory courses inside the school alongside the regular curriculum. They need help, and the help they want varies enormously. Some want teachers. Some want books. Some want dashboards and national rankings so parents can see where their child stands.

Selling through schools rather than straight to parents is a quieter model than the consumer edtech playbook, and it has one big advantage: the school already has the student's time. That same insight runs through ConveGenius's personalised learning for Bharat, which also bet that the classroom, not the app store, is where adaptive learning reaches scale.

Why 80% of the Schools Are in Small Towns

This is the number in the conversation that should make people stop. More than 80% of SpeEdLabs' schools are in small towns, semi-urban areas, district headquarters and villages. Metros and tier 1 cities are served too, but they're the minority.

Vivek's explanation is simple, and it's the kind of thing you only say with conviction if you grew up in Kasganj.

JEE doesn't grade on a curve for your postcode. A student in a district town is sitting the same paper as a student in South Delhi, usually with far fewer resources behind them. That makes structured practice, good analytics and a national ranking disproportionately valuable exactly where they've historically been hardest to get. And the same logic, he says, is now spreading to skills — robotics, coding and AI aspirations are growing fast in small towns too.

AI Deserves a Real Slot in the Timetable

STEMLearn.AI is the umbrella brand for all of SpeEdLabs' tech courses: robotics, coding, artificial intelligence, computational thinking and hackathons. It recently launched a national AI Olympiad. But the part of this section Vivek is most passionate about isn't a product. It's a scheduling argument.

Anyone who went to an Indian school knows exactly what he means. The subject that gets one period on a Friday afternoon is the subject everyone treats as optional, including the teacher. Vivek expects the next five years to be transformational, with every school — the big national chains and the small independent ones alike — forced to work out how many AI periods fit into a fixed weekly timetable.

India, China and South Korea

Asked whether Indian education lags its Asian peers, Vivek gives a balanced answer rather than a doom-laden one. Indian education, he says, has delivered a lot given the resource constraints the country operates under. The real problems are infrastructure, teacher availability and teacher skilling, not the system's basic intent.

He also points out that China, Japan and South Korea are every bit as competitive and pressure-heavy as India. They're populous, everyone wants the top college, and each has its own brutal entrance exam culture. The difference he sees is emphasis.

He adds an observation parents will recognise: today's students are simply different. Attention spans, aspirations and exposure have all shifted, and parents are willing to spend far more on education than they were twenty years ago. That's a tailwind, provided the product actually delivers outcomes.

Taking AI to One Lakh Government School Students

The most significant recent step is a partnership with the Uttar Pradesh Skill Development Mission. STEMLearn.AI signed an MoU with UPSDM in August 2026 to bring AI and robotics skilling to government school students in the state, starting by taking the AI Olympiad to more than one lakh students in Grades 9 to 12.

Vivek frames it as a starting point, not a finish line. If the outcomes hold up, he wants to scale it further within UP and replicate the model with other states. The reasoning is as much about economics as ideals.

This is a well-worn but important truth about Indian education. Private schools can buy programmes. Government schools, which educate the majority of the country's children, mostly can't — unless a state mission or a CSR budget underwrites it. Connected Technologies' work on government school digitisation ran into the same reality: the content can be ready, but reach depends on the state deciding to pay for it.

What's Next: IPMAT, Techfest and Financial Literacy

SpeEdLabs' expansion strategy follows the same thread it always has — aspiration. Where students are already motivated to work towards something, there's room for a product that helps them get there.

The financial literacy push is the one that feels most personal. A founder who spent eleven years in private equity and structured debt, now teaching teenagers how money works, is about as on-brand as a roadmap gets.

The Honest Version of the Edtech Winter

This is where the conversation turns, and where Vivek is at his most useful to other founders. He doesn't pretend the edtech boom and bust happened to other people.

His read of the sector is blunt. India has more than 25 crore students, the largest student population of any country, and COVID turned that into what looked like an enormous technology opportunity overnight. Then, in hindsight, a lot went wrong at once.

When the companies that raised the most failed to meet investor expectations, sentiment towards the entire sector soured. Everyone else paid for it. Vivek is clear that SpeEdLabs wasn't somehow insulated.

He doesn't dwell on how bad it got in this conversation, though elsewhere he has spoken publicly about how close the company came to shutting down. Here he focuses on what changed afterwards, and the numbers speak for themselves.

He's also more optimistic about the funding climate than the headlines. The problem of educating 25 crore students is too large to be solved by one business model, so capital will keep flowing to different approaches. There have been successful edtech IPOs, including some at billion-dollar-plus valuations, and private funding is returning. The difference is that everyone — founders and investors — is now far more careful about how money gets spent, how experiments get run, and how platforms get monetised.

A Founder's First Duty Is to Survive

Vivek's closing advice to founders is some of the clearest this podcast has heard, and it applies well outside education.

It's worth sitting with the first rule. It sounds defensive, almost unambitious. But it's the lesson of the entire edtech cycle in one sentence. The companies that survived got to learn from the boom. The ones that didn't just became case studies.

Key Takeaways

• Personalisation is about the right difficulty, not more content: Keeping each student in their "proximal zone" — stretched but not crushed — is what builds the practice habit that actually improves outcomes.
• Be honest about what's AI and what isn't: SpeEdLabs' personalisation runs mainly on algorithms and classical ML, with generative AI on the content side. Precision about your own stack builds credibility.
• The moat is pedagogy, not the model: The proprietary asset is how experienced teachers decided questions should be sequenced. The technology is replaceable; that judgement isn't.
• Aspiration is national, and so is the market: Competitive exams don't care about postcodes, which is why more than 80% of SpeEdLabs' schools are in small towns.
• Subjects need timetable slots to be taken seriously: One AI lecture a week signals that AI is optional. A permanent place in the timetable signals that it matters.
• Government school reach needs government or CSR money: The content can be ready, but affordable scale for public schools depends on state partnerships like UPSDM.
• Own your share of the boom's mistakes: Overfunding and a business model without fit hurt SpeEdLabs too. Changing the model to match real customer needs is what produced 6x growth.
• Survive first, then optimise: Survival buys the time to course-correct. Product-market fit and unit economics are the two numbers that decide whether you get that time.

About the Guest

Vivek Varshney is the Founder of SpeEdLabs. Raised in Kasganj in western Uttar Pradesh, he studied mechanical engineering at IIT Kanpur, taught at Career Point in Kota, and earned an MBA in finance and strategy from IIM Lucknow. He then spent about 11 years in finance and private equity with Morgan Stanley, Deutsche Bank and ICICI Venture before leaving to start SpeEdLabs in 2016.

SpeEdLabs is an adaptive, personalised learning platform for school students preparing for competitive exams including JEE, NEET and IPMAT. It works through partner schools, providing books, student and teacher apps, teacher training, national-level assessments, analytics and robotics and AI kits. Around 1.5 lakh students across roughly 700 schools use its products, with more than 80% of those schools in small towns. Its STEM brand, STEMLearn.AI, runs robotics, coding, computational thinking and AI programmes, a national AI Olympiad and an AI competition with Techfest, IIT Bombay. In August 2026, STEMLearn.AI signed an MoU with the Uttar Pradesh Skill Development Mission to bring AI skilling to more than one lakh government school students in the state.