Llamaroo is the AI lesson builder for primary schools. A teacher uploads a worksheet, a topic, or a voice note. Llamaroo writes a story-driven lesson with arcade-style games built around the same learning objectives, ready to play in class the next morning. Children join with a class code and a PIN. AI never speaks to a child.
Llamaroo Ltd · Companies House #17189912 · Incorporated 29 April 2026 · 71-75 Shelton Street, London WC2H 9JQ.
A teacher uploads a worksheet, slides, or a thirty-second voice note while making dinner. Five minutes later, Llamaroo has written a complete lesson built around their objectives: a story, the practice questions, and a game. Like Duolingo, but tied to whatever the teacher needs to teach this week.
Children join with a class code and a four-digit PIN. No accounts, no email, no AI chat. They follow characters through stories, answer questions to earn coins, then spend coins on arcade-style game breaks. Designed by ex-Candy Crush developers; engineered to keep the lesson and the game tied to the same learning objective.
OneRoster import provisions an entire school in one upload. The school's logo and colours render on every screen. An approved-domain policy locks staff invites. Public, linkable trust center the head, principal, or DPO can read in a morning.
There are two product categories teachers already use. AI tools, MagicSchool, Brisk, Diffit, SchoolAI, output a Word doc. Game tools, Kahoot, Blooket, Gimkit, output a quiz with no link to the lesson. Llamaroo is the first product to do both, generated from the same teacher input: a complete, curriculum-aligned lesson and the game children actually play, in one workflow, in five minutes. Reach Capital made the same observation when they invested in Curipod and noted it "does not tout itself to be an AI company." We extend that thesis: Curipod made interactive slides; Llamaroo makes the full Duolingo-style experience around the teacher's existing objectives.
The unit of distribution is the classroom, not the user. Each lesson surfaces Llamaroo to twenty-five to thirty children at once. Teachers find us on the free plan, use it, and word travels in two directions: home, through siblings and parents; and across the staff room. A head, principal, or director of learning then asks for School Pro or Enterprise to bring the staff under one invoice and one policy. This is the loop Nearpod ran on its way to 75% of US classrooms before Renaissance acquired it for $650M, and the loop MagicSchool ran to two million educators in under twelve months.
Rohit Kossery is a Morgan Stanley VP, ex-Goldman Sachs, with 12+ years in asset management. He is a CFA charterholder, ACCA-qualified, and holds an Oxford MBA. Rohit has worked with Ujji, a seed-stage gamification startup for motivational training, and KPB Family Office, a seed-stage VC led by veteran investor KP Balraj. He has mentored students and young professionals since 2014 and believes deeply in education as a leveller. As a lifelong gamer, Llamaroo brings together his finance background, teaching experience, and belief in games as a powerful engagement tool. LinkedIn.
James Effarah is Head of Product at Odicci, a gamification startup for enterprise retailers, where he grew ARR from £500K to £1.2M in 12 months. He has 10+ years of experience across product, growth, and distribution spanning AI, gaming, SaaS, and retail. He worked on Candy Crush at Activision Blizzard King and led growth at Upgrow, a San Francisco agency for VC-backed SaaS and AI startups, driving $6M+ in incremental revenue for clients including Prisma, InfluxData, PingCAP, Materialize, and Deepset AI. James holds an Oxford MBA and has mentored 120+ aspiring product managers through MentorCruise and Oxford, applying game design principles to education and career development. LinkedIn.
Both founders are Oxford MBAs. The partnership runs through the Saïd Business School as a recurring product workshop with current students and alumni. We are scaling to three to five institutional design partners by June 2026.
Stories and lyrics stick after one pass. Course material fades even after weeks of study. Why is that? It is the question that drove us to Llamaroo. Songs reward the brain with story, repetition, and play; school content rarely does. The design partner programme exists to test that thesis on real students, in real teaching contexts, with the educators who set the curriculum. The University of Oxford, through Saïd Business School, is our first institutional design partner. We are scaling to three to five institutional partners by June 2026.
Each engagement is a low-risk pilot. One internal sponsor on the partner side, one teaching context, one interactive module, opt-in students. We provide the build and the iteration; the partner provides the classroom and the feedback. Small enough to ship in weeks, real enough to stress-test the format on actual learners.
One case or concept. One interactive module. Opt-in student participation. Small enough to ship in weeks, real enough to put in front of a live cohort.
Course design guidance from the founders. A partner-themed account with branded lessons. Iteration after every cycle of feedback, so the product evolves around how the partner actually teaches.
One internal sponsor inside the institution. One teaching context where the module fits. Staff and student feedback after the cohort plays. That feedback shapes the next iteration.
Llamaroo Ltd is a UK private limited company. This is the company's first priced round. Round size, valuation, and instrument are shared with qualifying recipients on request and not published on this page.
The questions an SEIS investor or angel actually asks. Answered directly.
An AI lesson builder for primary teachers. Teachers pay individually for Teacher Pro; schools pay for School Pro and Enterprise on a per-school invoice.
A teacher uploads a worksheet, slides, or a voice note. Llamaroo writes a complete, story-driven lesson with arcade-style games built around the same learning objectives, in five minutes. Children play it from a class code and a PIN, with no account and no AI chat. The free plan brings teachers in. Teacher Pro upgrades them. School Pro and Enterprise sell the school layer around them: OneRoster provisioning, Brand Kit, approved-domain security, classroom directory, and one invoice for the staff. The product is in use today, on real classrooms, with 250+ teachers across the US and UK.
AI tools give a teacher a Word doc. Game tools give a teacher a quiz that has nothing to do with the lesson. Llamaroo gives a teacher both, generated from one input.
The category split today is real. MagicSchool, Brisk, SchoolAI, Diffit, and Eduaide ship documents. Kahoot, Blooket, and Gimkit ship quizzes that float free of the lesson. Teachers stitch the two together by hand. Llamaroo collapses both sides into one workflow and one runtime: the lesson and the game are generated from the same input, against the same objectives, and play together. Reach Capital wrote the same investment thesis with Curipod and noted it "does not tout itself to be an AI company." Curipod made interactive slides; we make the full Duolingo-style experience around the teacher's existing curriculum.
Because that is the architecture a head, principal, or DPO will sign. It is also where the money is.
AI runs only on the teacher's authoring path. Children play a sealed, deterministic experience. This is precisely what the DfE Generative AI Product Safety Standards, KCSIE 2025, the ICO Children's Code, and the US Department of Education's 2023 AI guidance reward. Selling AI directly into the hands of children also limits the buyer to parents one at a time. Selling the AI to teachers makes the school the buyer; an entire staff in one purchase order, on a much longer customer lifetime.
Each lesson surfaces Llamaroo to twenty-five to thirty children at once. The classroom is the unit. Comparable: Nearpod, $650M acquisition by Renaissance.
Teachers find us on the free plan, use it, and word spreads in two directions: home, through siblings and parents, and across the staff room. A head, principal, or director of learning then brings in School Pro or Enterprise. This is the same loop Nearpod ran to 75% of US classrooms before its $650M sale to Renaissance, and the loop MagicSchool ran to two million educators in under twelve months. Sales-cycle benchmarks per Novel Capital: individual teachers one to three weeks; school admin two to three months; multi-academy trusts and US districts six to twelve months.
TAM ~$340B (global K-12 EdTech by 2030, HolonIQ). SAM ~£900M (UK and US primary at our list pricing). SOM ~£20-30M ARR at 5% teacher and 1% school penetration.
Bottom-up: ~218,500 primary teachers across ~16,800 UK schools (DfE, 2024) and ~1.5M elementary teachers across ~85,000 US elementary schools (NCES, 2024). Combined ~1.7M teachers, ~100K schools. At Teacher Pro and School Pro list pricing the SAM is approximately £900M ARR. The £20-30M ARR five-year line is not the headline opportunity; it is the quiet, geography-light path to a Series A. The headline outcome is a Nearpod or MagicSchool-shape distribution with a similar acquisition multiple.
AA drafted, pipeline first, then file. By month 24: a teacher base in the tens of thousands, three to five paying institutional partners, AA granted, runway through Series A.
Per HMRC VCM60130, an Advance Assurance application from a company that has not previously raised under tax-advantaged schemes must include named prospective investors, addresses, and intended amounts. We are completing that pipeline before filing. AA will be in place before first share issue. By month 24 the plan is: tens of thousands of active teachers; three to five paying institutional design partners and two to three paying multi-academy trusts or US districts; AA granted; Cyber Essentials and DPIA published; runway through a Series A from a US ed-tech specialist (Owl, Reach, Bessemer, or Bain Capital Ventures).
