Arivihan Raises $10M Series A to Scale Rs 51 AI Tutoring Plan

A $10 million bet on India's mass-market students

Arivihan, an edtech startup based in Indore, has raised a $10 million Series A. The round came from Accel and Prosus Ventures, both of which already held stakes, and the company says its total funding now exceeds $15 million.11 Existing GSF-linked angel investors added about $200,000, so some outlets report the round as $10.2 million. The two figures describe the same deal, not competing accounts.1219 The company raised $4.17 million in a pre-Series A round in July 2025, and it got its seed money through Accel's Atoms accelerator in March 2024.12

The company has not announced a valuation. Regulatory filings reported in the days before the announcement put the post-money figure at about ₹570 crore, or roughly $60 million. That is about 3.3 times the ₹171 crore valuation of the pre-Series A round.1714 The same filings show Accel and Prosus each putting in about ₹47.48 crore. After the round, Accel India holds roughly 23.54% and Prosus about 18.57%.17

The money is meant to take Arivihan into new states and further into the CBSE curriculum. It will also fund AI research, support for more Indian languages, performance marketing and local distribution networks.12 Co-founder and CEO Ritesh Singh Chandel has said he wants to cover almost all state boards by the end of the year and expand the company's preparation for NEET, India's medical entrance exam.14 The company also plans to strengthen its position in Madhya Pradesh, Uttar Pradesh, Rajasthan and Bihar and to launch crash courses.13

The ₹51 pivot behind the numbers

The most notable detail in the coverage is a pricing change. Arivihan used to charge about ₹3,000 a year on average for board-exam preparation and had about 25,000 paying students last year. After it introduced a ₹51 annual entry plan, Chandel says 4 to 4.5 lakh students (400,000 to 450,000) signed up this year.11 He describes the aim plainly: the company wanted volume. Referrals brought in 18% of the ₹51 subscriptions.11

This growth in user numbers should be read with care. Arivihan has not disclosed its revenue or how many students moved up to its more expensive plans. A sixteen-fold rise in paying users at a fraction of the old price does not mean revenue grew anywhere near as fast.11 The most recent audited figures are for FY25: revenue of ₹3.14 crore, up 14.8 times from ₹21.29 lakh in FY24, and a net loss of ₹4.45 crore. FY26 accounts have not been filed yet.1715 Taken at face value, the ₹570 crore valuation is more than 180 times trailing FY25 revenue, though that number is out of date given how fast the company is growing.17

The valuation looks rich, and the fact that both leads were already investors makes it hard to treat the price as a clean market signal. In effect, Accel and Prosus are marking up a position they already hold.17 One sympathetic reading is that insiders can see the retention data, and their conviction is itself information. Prosus investor Dhruv Gupta said the company had grown quickly while keeping student engagement high. He presented it as proof that AI can deliver individualised learning at mass-market prices.12 The other reading is that the round prices in a monetisation story the public numbers do not yet support. Whether the company can move a large share of ₹51 users onto higher plans is the central uncertainty, and one analysis of the round names that conversion as the key long-term risk.20

How much of this is really AI tutoring?

Coverage of what Arivihan actually sells varies a lot. One education outlet calls it a "teacherless" platform that replaces live teachers with AI.18 Another calls it automated, one-on-one personalised tutoring.20 Chandel's own description is more modest. The current classes mix human teachers with AI: one teacher may teach a live class of up to 5,000 students, while AI answers each student's questions using that student's mistakes and practice history. A true one-on-one AI tutor is still on the roadmap and does not exist yet.11

The distinction matters, and the founder's version should be the one that counts. Arivihan is a hybrid. It uses AI to handle individual questions so that one human teacher can lecture thousands of students. That is easier to defend than replacing teachers outright, and it fits the research discussed below better.

The company's market is clear. It started with Class 12 board-exam preparation in Madhya Pradesh and later added other state boards and NEET. It says about 80% of subscribers live in Tier 3 cities or rural areas.11 The product was built for Hindi-speaking students first, Marathi is in testing, and lectures can be downloaded for students with patchy internet.11

Coverage also disagrees on basic details. Some reports place the company in Bengaluru, while most describe it as Indore-based.1511 Its founding year appears as 2022, 2021 and 2020 in different reports.112014 None of this changes the deal, but it shows how much of the coverage is built on press-release material that was not checked.

The outcomes claims, and what the research says

Arivihan points to some results. About 5,000 students used its NEET preparation in the latest exam cycle, and three of them reached top-100 ranks. Internal data shows that 75% of students who completed five mock tests raised their marks by 25% within 30 days.11 However, the company did not say how many students were in that group or compare them with students who didn't use the platform.11 Students who choose to take five mock tests are probably more motivated than average, so the figure says little about what the platform itself causes. One review of the round lists what would settle the question: learning outcomes measured against a clear comparison group, retention, error rates on difficult questions, and how often students need to be handed to a human.19

The research on AI tutoring is mixed, and the mix is informative. The best-known positive result is a Harvard randomised controlled trial in an undergraduate physics course with 194 students. Students using a custom AI tutor at home had median learning gains more than double those from an in-class active-learning lesson, and they spent less time on the material.21 The authors estimate the effect at 0.73 to 1.3 standard deviations. They stress that the tutor was carefully designed around research-based teaching practices.21 The study is also short: it was a two-week crossover, not a year-long study.24

Trials in lower-income settings are closer to Arivihan's market. In a World Bank-linked after-school pilot in Nigeria, where teachers guided the sessions, scores rose by 0.31 standard deviations. In Ghana, the Rori math tutor produced an effect size of 0.36 over about eight months.25 A trial of the Medly platform among English GCSE students found an effect of 0.33, though the vendor paid for the study and 30.7% of participants dropped out.25

The negative results are just as relevant. In a Turkish high school, students using an ordinary ChatGPT-style interface did 17% worse once the AI was taken away. A version that gave teacher-designed hints instead of answers was statistically no different from no AI at all.25 At the University of Maryland, students given a course-specific GPT-4o study assistant ended with final grades 0.37 standard deviations lower in the same-course sample.25 Stanford's Tutor CoPilot, which supports human tutors rather than replacing them, raised short-term mastery by 4 percentage points but did not improve end-of-year test scores.25

Taken together, design matters more than whether AI is present. Systems that give students the answer can raise practice scores while weakening independent learning. Systems that guide students through reasoning, with a human involved, do better.2228 One education outlet's summary of the research is that these tools work best when they support teachers, not replace them.18 By that standard, Arivihan's current hybrid model is on the better-supported side of the evidence. Its planned fully automated tutor would be on less certain ground, especially for students preparing for high-stakes exams, where one wrong worked solution can cost their trust.20

A selective market rewards a specific thesis

The round stands out because Indian edtech money is so scarce now. Tracxn reports that equity funding in the sector fell from about $4.3 billion in 2021 to $214 million in the first eight months of 2026, and the number of funded rounds dropped from 368 to 36.12 The median round size has gone the other way, reaching $1.1 million this year, about twice its level in any of the previous five years.5 At $10 million, Arivihan's round equals almost 5% of all equity money the sector raised in the first eight months of 2026, by that count.

The sector's history explains why investors are cautious. BYJU'S raised the most of any Indian edtech company, $4.6 billion, but its parent has been in insolvency proceedings since July 2024. Unacademy was absorbed by upGrad in an all-stock deal.1 Physics Wallah raised far less than either, yet it listed at a $3.6 billion market capitalisation, and it has been expanding its offline centres quickly.27 K-12 still attracts the largest share of sector funding, 51% over the 2021–2026 period.1

Arivihan sits between these models. It is not trying to match Physics Wallah's offline footprint. It is betting that very low prices, vernacular content and AI-assisted scale can reach students the big coaching companies never served profitably. Funding that bet makes sense in a market that now rewards working business models over growth at any cost. But there is little public evidence for it so far: no disclosed revenue since FY25, no published conversion rates, and learning claims without a comparison group. The FY26 filings and any independent outcomes data will show whether the ₹570 crore valuation was justified or premature.17