The problem
The barrier wasn’t access to content. It was children who had learned to believe they were “behind”.
An offline-first learning platform for underserved children in India, where the AI supports without ever judging.
The barrier wasn’t access to content. It was children who had learned to believe they were “behind”.
Thirteen interviews, then a narrated, offline-first app where AI adapts quietly in the background.
All 25 high-fidelity task runs succeeded with no assistance. Emotional response was positive on every task.
01The problem
India runs one of the world’s largest public education systems, yet many children in government schools struggle with basic literacy and numeracy. Mainstream EdTech doesn’t reach them, and when it does, it often makes things worse.
Leaderboards, visible grades and instant failure feedback replicate the classroom judgement these children already fear.
What underserved children need is not more content, but a learning environment that proves, through every interaction, that they are capable.
02Discover
Ethical constraints ruled out direct research with minors. Insight came instead from adults with years on the ground: NGO educators, community facilitators and child-rights practitioners, plus experts in child rights, cybersecurity and pedagogy.
Every insight had to hold on both sides: what the people beside these children said, and what the secondary research showed.
01Access constraints
“Most homes have one mobile phone, but children don’t have personal access.”
Learning collapses without offline-first, low-bandwidth, interruption-proof design.
ASER 2023: uneven device access, shared phones. UNICEF and World Bank 2022: poor connectivity in rural and tribal India.
02Motivation needs safety
“Curiosity is their default. They want to learn more.”
Motivation is high, but it needs emotionally safe, structured, guided experiences.
Gamification improves engagement for underserved learners. Emotional safety is key for low-literate learners.
03Visual, narrated, local
“Story-based lessons in regional language work best.”
Understanding depends on multimodal, localised, narrated content, not text.
ASER 2023: low reading proficiency. Multimodal learning boosts retention; local language improves comprehension.
04EdTech misaligns
“Too much to click equals confusion.”
Mainstream EdTech excludes low-resource children; a new design grammar is needed.
Commercial EdTech assumes literacy, stable internet and parental supervision.
05Home disrupts continuity
“Children juggle home chores and education.”
Tools must assume noise, interruption and zero supervision.
UNICEF 2022: chores, hunger and caregiving disrupt learning.
06Teachers adopt for value
“Dashboards should simplify their role.”
Teachers adopt only when tools reduce effort and show visible impact.
NEP 2020: teachers are overburdened; tools must reduce workload.
Needs narration, reassurance and repetition.
One-path flows, never-failing navigation, warm voice guidance.
Avoids long tasks, fears mistakes, abandons when confused.
Short lessons, proactive explanations, confidence-first feedback.

Overloaded and time-poor. Adopts only if impact is immediate.
Pre-built lesson kits, minimal dashboards, zero training burden.
Learns in bursts amid interruptions.
Auto-save, pass-the-phone mode, offline continuity.
03Define
No punishment and no red marks, anywhere.
AI adapts silently, without exposing the child’s struggle.
Difficulty is reframed, never labelled as failure.
Recognition for trying, not for scoring.
04Develop
No feature may increase emotional or cognitive load for a novice learner.
Must have23
Should have
Gentle motivation, progress without comparison
Could have
Limited, low-pressure peer support
Won’t have, for now
Anything that compares or exposes a child
Against the landscape, Smart Slate takes a space of its own: the structure of a curriculum platform with the low cognitive load of something far simpler.
05Silent AI
Most EdTech puts its AI on show through dashboards and rankings. Smart Slate hides it from the learner entirely. Three mechanisms do the work.
01
The system notices hesitation, narration repeats and backward navigation. It doesn’t jump to an “easier version”. The app simply becomes gentler: shorter lessons, simpler narration, earlier hints.
02
If a child struggles with number sequencing but responds to stories, the next explanation arrives as a story. Confidence first, then back to the abstract concept.
03
Practice is offered as optional and repeatable, with no marks or grades, and pitched so success is likely.
01
Before teaching, gauge where the child is.
02
Explain, re-explain, then check gently.
03
Close the loop, then plan the next lesson.
Decision points, where the lesson quietly changes course. The child only ever sees the next screen.
Where the adapting shows
The supervisor view lists what the system observed and how it adapted: frequent pauses on multi-step division, retries on vocabulary matching, lessons broken into smaller steps, hints held back fifteen seconds to encourage recall.

On-device models, multilingual text-to-speech and batch sync keep all of this working offline, and mean personalisation never depends on constant cloud monitoring.
06Deliver
Visual-first communication, large touch targets, predictable layouts and micro-lessons of two to five minutes. Every instruction can be heard, in English or Kannada.

















Each flow plays through the prototype. Tap a step to jump to it.
Home
The dashboard greets the learner by name and leads with Continue Learning. Start a Lesson and Complete Homework sit below, then activity and the library.
Why this matters. Testing reduced the dashboard to one clear next step, so a hesitant child never has to choose between parallel actions.




07Testing & iteration
Low-fidelity testing with proxy participants surfaced a problem I hadn’t designed for. “Learn with Friends” had no competitive mechanics, yet testers kept asking whether it involved competition or being watched. Two of four attempts failed.
Found
Peer interaction introduced performance anxiety, especially for lower-confidence users. Multiple parallel actions caused hesitation.
Changed
Peer learning left the core flow and became asynchronous notes sharing. Quizzes became “Homework”. The dashboard was reduced to one obvious next action.
Five proxy participants, two educators and three teaching assistants, completed five tasks. Each was scored on time, errors and assistance, then rated for how it felt.
| Task | Avg. time | Errors, total | Emotional response |
|---|---|---|---|
| Start today’s learning | 36s | 2 | Positive |
| Seek help while learning | 38s | 3 | Positive |
| Find notes and homework | 37s | 3 | Positive |
| Use Word Help | 37s | 3 | Very positive |
| Use AI-aided support | 40s | 3 | Positive |
Word Help drew the warmest response: “reassuring” and “helpful without pressure”. The AI guide scored positive but more cautiously, which supported keeping it a bounded, intent-based helper and not an open chatbot.
08Accessibility audit
I audited the prototype against WCAG-aligned mobile principles and child–computer interaction guidance.
Risk
Icons without labels push low-literacy learners to memorise.
Design action
Pair every critical icon with a short label; tap-and-hold reads it aloud.
Risk
Soft brand colours can wash out on budget devices or outdoors.
Design action
Enforce minimum contrast for text, buttons and navigation.
Risk
Children tap imprecisely; packed controls cause accidental taps.
Design action
Large targets, more spacing, no small inline links.
Risk
Narration is the main access route, so hidden controls frustrate.
Design action
Persistent play, pause and repeat. “Repeat” is a primary action.
Risk
Unclear what works without a connection.
Design action
An offline banner, a “Saved for Offline” section and plain-language sync messages.
09Critical reflection
For low-resource learners, the most meaningful use of AI is often invisible.
Keep exploring
A satellite-enabled wearable that keeps vulnerable people reachable when networks fail.
2025 – 2026Read case study →
I tested the “accessible default” with 73 people, then designed a system that doesn’t depend on it.
2026Read case study →
A heuristic evaluation of the document wallet used by 550+ million citizens.
10 findingsRead case study →