Pre-launch · Built in India
On Edukul, children build games, stories, experiments and plans of their own. A bounded AI guide asks questions and offers small hints instead of handing over answers, and parents can see which work their child did independently.
Edukul is a creation-based learning platform for children aged 3–17, for families and schools. We're in development, India first; join the waitlist for early access.
AI tools are now within reach of many children. The skill worth learning is not avoiding them but judging and directing them: asking a good question, checking an answer, and explaining a decision in their own words. Today, few children get to practise that.
A general chatbot will happily write a child's homework. The work gets done, but the child skips the part where learning happens: trying, getting it wrong and working out why.
Exams and worksheets tend to measure what a child remembers. Planning, testing an idea and defending a choice are harder to mark, so they get less time.
Much of what children do on screens is watching and scrolling. Very little of it ends with something they made and can explain.
Weighing evidence, improving a first attempt and saying why are learned by doing them often, with a patient mentor nearby.
In a gurukul, children learned alongside a mentor, through practice, questions, reflection and real tasks. Edukul borrows that teaching approach and puts it into software, with a bounded AI guide and responsible adults in the mentor's place. It is a teaching approach, not religious content: no observance or belief is required, and there is no caste or gender hierarchy.
A bounded AI guide picks up where the child left off, with parents and teachers overseeing. It is designed to avoid cues that make a child emotionally dependent on it.
The guide asks for reasons, compares explanations and invites the child to challenge an assumption, including its own.
Repeated practice in context, followed by a short reflection. Attempts and reflections are recorded separately.
Missions at home, in nature, in the community and at the maker's table carry the work off the screen.
Children choose goals, plan their time and explain which strategy helped them.
Optional projects that help people or the environment, shared only with consent.
Evidence across social, practical and thinking skills. Never a single intelligence ranking.
The child picks a challenge suited to their age band and goals.
They make something real: a game, a story, a plan, a model or an experiment.
They run it and see what actually happens, not what they hoped would happen.
They change one thing and test again. The AI guide asks questions; it does not make the fix.
They say or write why they made their choices, in their own words.
The work and the explanation are saved as evidence, marked as done independently or with help.
Picks the café challenge: keep a small café running for a week of virtual customers.
Sets prices and a daily order of samosas and lemonade in a cost sheet.
Runs the simulation. Lemonade sells out by Tuesday, and twelve samosas are thrown away.
"What was different on the days with waste and the days without? What could you change first?"
Lowers the samosa order, raises the lemonade order, and tests again. Waste drops.
Writes in the café journal why the lemonade price stayed the same and the samosa order went down.
Saves the cost sheet, simulation and journal. The pricing explanation is recorded as independent work.
| Tutoring classes | General AI chatbots | Edukul | |
|---|---|---|---|
| Who does the work | The child, following the tutor's plan | Often the chatbot | The child; the AI guide asks questions and gives small hints |
| What gets saved | Worksheets and test marks | A chat history | What the child made, how they explained it, and whether they did it independently or with help |
| How a parent can tell the child did the thinking | Marks and the tutor's feedback | Hard to tell | Independent attempts are recorded separately from assisted ones |
| Adult control | Schedule agreed with the tutor | Mostly built for adults, with few controls for children | Parents and teachers set goals and daily limits |
| Ages covered | Usually one subject or grade range | Generally not designed for young children | One approach from 3 to 17, designed separately for each age band |
Ages 3–5
Adult-led play with voice and large pictures, and no reading needed. Children notice, sort, describe and take turns.
Ages 6–8
Guided voice and drawing activities with an adult nearby: asking questions, explaining a choice and finishing a mission at home.
Ages 9–12
Comparing evidence, solving problems, working within a budget and improving a prototype over several sessions.
Ages 13–15
Researching, designing, building, testing and presenting a responsible solution, using AI under supervision and checking what it says.
Ages 16–17
Complex, independent projects and real-world problem-solving with AI as a collaborator. The supervision model for this band is being designed and reviewed.
A parent or guardian creates the child's profile and gives consent before the child can use Edukul or the AI guide. Consent can be withdrawn at any time.
No advertising, no endless feeds, no public profiles and no messaging with strangers.
Each age band has a conservative default daily limit. Parents can lower it; raising it takes a deliberate adult action within set ceilings. Off-screen missions don't count as screen time.
The guide works inside reviewed learning journeys, tells children it is a computer helper that can make mistakes, and is designed to avoid emotional-dependency cues. What children write is screened, and anything concerning is flagged and logged.
Each learning journey is written by one person and approved by a different reviewer before children can see it.
Children's learning data is designed to be stored in India, in AWS's Mumbai region. Parents can export or delete their child's data.
Privacy is designed around India's Digital Personal Data Protection Act, 2023: a clear purpose, verifiable parental consent, minimal data and deletion on request.
Edukul is pre-launch. This is a dated log of what exists today in our development environment, tested with synthetic data only. No children use it yet.
Guardian accounts, consent records with withdrawal, a tamper-evident audit log, and per-child data export and deletion.
Staff write learning journeys in a studio. A different reviewer must approve each one before it can be published.
Children build programs with coding blocks in a sandboxed preview, and the server replays each run to confirm what the program did.
The child's app walks through plan, build, test, explain and save, with an AI guide that gives hints and a portfolio that saves as they work.
Parents see independent and assisted evidence side by side, set limits, and export or delete a child's data.
One playable journey per band, from a garden helper for ages 3–5 to an idea-to-prototype project for ages 16–17.
Schools, classes and rosters, school consent with parent opt-out, child sign-in on school devices, and teacher assignments and review are in place. The remaining pilot features are under way.
Approved missions that continue a project away from the screen, with adult confirmation. The first round is built; more formats are being added.
Staging and production environments in India, safeguarding and legal reviews, then a pilot with families and schools. No launch date is set yet.
Founder
Founder of Edukul, building it from Jaipur. Responsible for product and engineering.
LinkedIn: Jitin Gambhir (opens in a new tab)Not yet. Edukul is in development and pre-launch, starting in India. We haven't set a launch date. Join the waitlist and we'll email you when early access opens.
It asks questions, gives small hints when asked and points out what a test showed. It is designed not to do the child's work, not to hand over final answers during independent practice and not to act as a friend. It tells children it is a computer helper and can make mistakes.
Default daily limits are 15, 25, 35 and 45 minutes for ages 3–5, 6–8, 9–12 and 13–15; the limit for 16–17 will be set before launch. Parents can lower any limit. Off-screen missions are part of the design and don't count as screen time. The defaults are conservative starting points to be tested in pilots, not medical guidance.
Children's learning data is designed to be stored in India, in AWS's Mumbai region, and parents can export or delete it. This website's waitlist is separate: it holds adult contact details only, is processed by Cloudflare and may be stored outside India. The privacy policy has the details.
No. Gurukul names a teaching approach: learning with a mentor, through practice, reflection and real tasks. There is no religious content or observance, and no caste or gender hierarchy.
Pricing hasn't been decided. Joining the waitlist is free, asks for no payment details and doesn't commit you to anything.
Schools are one of our two audiences. We're preparing pilots with classes, teacher review and school-device sign-in. Choose "For schools" on the waitlist form to register your interest.
Children aged 3 to 17, in five bands. Each band is designed separately: voice and pictures with an adult for ages 3–5, through to independent projects for ages 16–17.
Tell us whether you're a parent or a school. We'll use your details only to contact you about Edukul's early access and pilots.