In the crowded arena of AI-driven education tools, where chatbots spit out instant answers to homework woes, Fermi.ai emerges with a contrarian pitch: make learning deliberately harder. Launched this week by Peeyush Ranjan, a former general manager and vice president at Google who also served as chief technology officer at Walmart-owned Flipkart and vice president of engineering at Airbnb, the Singapore-headquartered startup targets high-school students grappling with math, physics and chemistry. Available for free initially via subsidiaries in the U.S. and India, Fermi.ai eschews quick fixes in favor of step-by-step guidance that mimics a teacher’s probing questions.

Ranjan, who joined Meraki Labs—a startup studio he co-founded with Myntra co-founder Mukesh Bansal—as a partner last year after his Google stint leading AI products like Google Assistant and parts of Gemini, positions Fermi.ai as a bulwark against superficial learning. ‘AI has made hard things easy for all of us. But as lifelong learners we saw that this process of making hard things easy could hamper learning, unless AI is used the right way,’ Ranjan told reporters in a press release distributed by PR Newswire. The platform, built entirely on multiple large language models with no human tutors involved, rolled out after a three-month pilot with 79 students across Bengaluru, North India and Silicon Valley, where low performers boosted scores from around 2 out of 10 to 6.7, hinting rates fell to 13-15%, and reliance on hints dropped 21% according to Entrepreneur India.

Ranjan’s Path from Big Tech to EdTech Pioneer

Ranjan’s pedigree lends credibility to Fermi.ai’s ambitions. A graduate of the Indian Institute of Technology Kharagpur, he cut his teeth as a software developer at Hewlett-Packard before ascending through engineering leadership at Flipkart, Airbnb and Google, where he oversaw payments, fintech and next-billion-users initiatives as detailed in his Crunchbase profile. At Meraki Labs, backed by Bansal’s personal capital from ventures like Cult.fit, Ranjan shifted focus to AI-native startups. ‘In a post-AI world, every industry has to rethink from first principles,’ he said in an interview with Business Today.

The timing aligns with edtech’s post-pandemic reckoning, especially in India ($8 billion market) and the U.S. ($5 billion), where Ranjan sees acute pain points in STEM mastery amid exam pressures like JEE, AP and IB, per Business Standard. With a lean 10-person tech team and no external funding yet, Fermi.ai plans to raise capital post-pricing trials, betting on outcomes over scale.

Four Pillars Defying the Answer-Machine Trend

Fermi.ai rests on four pillars: an adaptive real-time tutor that provides hints like ‘read the problem again, identify variables, draw a free-body diagram’ rather than solutions; a handwriting-first smart canvas for stylus-based equations and diagrams; a concept graph-driven question bank tied to curricula; and a diagnostics layer spotting reasoning gaps for students and teachers. ‘We built Fermi.ai to support thinking, not replace it, to protect the ‘productive struggle’ that leads to actual mastery,’ Ranjan emphasized to Entrepreneur India.

Unlike rivals handing final answers, Fermi.ai’s flows—Homework Assist, personalized practice and targeted revision—flag errors in real time and adapt prompts, fostering ‘concept ownership’ over guesswork. Educators gain classroom insights into ‘silent struggles’ and misconceptions, as noted in the Yahoo Finance reprint of the launch release. Bansal added: ‘We have entered an era where AI can solve any equation, but it can’t yet explain why a student’s logic faltered at step three. Fermi.ai isn’t here to give answers; it’s here to provide the map and the mirror.’

Pilot Proof and Market Pressures

The pilot’s 15,000+ concept tests showed consistent mastery gains, with poor starters improving markedly and overall engagement sustained. ‘What we’re taking away is the fear of getting stuck,’ Ranjan told Business Today. Abandonment stayed low at 13-15%, a stark contrast to fleeting chatbot interactions.

India’s value-conscious consumers demand competitive pricing, while U.S. parents seek outcomes amid ‘AI tutor race’ skepticism, as BriefGlance framed it: ‘The industry has spent years building AI that gives answers. We built Fermi.ai to do the opposite.’ Ranjan eyes expansion to biology for NEET, engineering, accounting and data science, plus ahead-of-school concepts, per MyStartupNews.

Backing and Broader AI EdTech Push

Meraki Labs provides seed support, with fundraising eyed in six months. Ranjan’s Silicon Valley base aids global ties, complementing Bansal’s Bengaluru operations. ‘Education is fundamental to society’s development, and AI is here to stay in students’ lives,’ Ranjan told MyStartupNews. X buzz from outlets like @ETtech and @businessline highlights the launch’s traction.

As edtech pivots from content access to interactive depth, Fermi.ai’s mastery-first model tests whether investors and users reward rigor over rapidity. With Ranjan’s track record scaling billions-user products, the startup enters a sector hungry for tools that build thinkers, not just test-takers.