

Singapore, January 23, 2026 – In an edtech landscape still licking its wounds from a brutal funding winter, serial entrepreneur Mukesh Bansal is back with a fresh swing. The man behind Myntra and Cult.fit has teamed up with ex-Google and Flipkart heavy-hitter Peeyush Ranjan to launch Fermi AI, a Singapore-headquartered startup that’s flipping the script on AI tutors. No more spoon-feeding answers like lazy chatbots – Fermi is all about guiding high schoolers through the gritty process of STEM mastery, one stylus scribble at a time. It’s the kind of thoughtful pivot that feels like a breath of fresh air in a sector desperate for real learning, not just quick fixes.
Bansal and Ranjan aren’t new to building winners. Fresh off their roles as partners at early-stage incubator Meraki Labs (which is backing Fermi with advisory muscle and seed cash), they’re targeting grades 9-12 in India with after-school coaching in physics, chemistry, and math. Stateside, it’s advanced placement (AP) classes for STEM subjects. Launching free for all as a pilot, the app’s full rollout is still cooking, but early buzz from Bengaluru, North India, and Silicon Valley pilots has educators nodding. “Fermi.ai isn’t here to give answers; it’s here to provide the map and the mirror – showing students how they think and giving teachers the visibility to lead them back to the path of Mastery,” Bansal says, nailing the philosophy in one punchy line.
What sets Fermi apart? Most AI edtech tools spit out solutions faster than you can say “ChatGPT,” turning kids into copy-paste pros with zero retention. Fermi flips that – it withholds final answers, walking students step-by-step through problem-solving like a patient coach spotting flawed logic at step three. The interface? Stylus-first magic on tablets or Chromebooks, mimicking pen-and-paper for better retention. Draw diagrams, jot equations, build reasoning – it’s interactive, not passive video drudge.
Powered by a mashup of top LLMs (OpenAI, Google, you name it), Fermi packs four core pillars: an adaptive real-time tutor, handwriting canvas, curriculum-linked concept graphs tied to exams like JEE, IB, AP, and diagnostic tools that expose weak spots. Teachers get analytics dashboards for targeted interventions, while schools can weave it into curricula. No creepy data training on student work – privacy first, with teacher uploads staying put.
This is day one, but the vision is massive. Post-pilot, Fermi eyes teaching brand-new concepts beyond homework drudgery, expanding to biology for NEET prep, engineering, accounting, data science, and more. Global push is priority one, with pilots scaling fast. Monetization? Per-student seating (teachers free), but they’re proving traction first – “serious promise qualitatively and quantitatively,” Ranjan notes. No external funding yet; Meraki’s got their back for now.
Timing couldn’t be spicier. Edtech funding cratered to an 8-year low of $249 Mn in 2025 (down 56% YoY), per reports, as investors shun bloated late-stagers for nimble AI plays. Video-era passivity? Done. Fermi bets on “productive struggle” – Ranjan calls out how AI shortcuts weaken understanding. “Education is fundamental… AI is here to stay, but if we don’t do it properly, it’ll be used negatively,” he warns. Past woes like content access are solved; now it’s about interactive depth.
Fermi AI isn’t just another edtech entrant – it’s a wake-up call for an industry at a crossroads. By prioritizing thinking over answers, Bansal and Ranjan are crafting a long-term learning engine that could redefine STEM education amid AI’s rise. In a funding-scarce world, their bootstrapped smarts and proven pedigrees position them to thrive where giants stumbled. If pilots deliver, Fermi won’t just disrupt – it’ll rebuild how kids (and teachers) tackle the future. Watch this space; the next big edtech wave might just be the one that actually teaches.