iSTART Early
Turning reading practice into a game kids want to finish.
An intelligent tutoring system that teaches 3rd and 4th graders reading strategies, with speech and language AI giving real-time feedback. It was built for high schoolers, and it showed: only 60% of learners stayed engaged.
Full case study coming soon. This is a short version while I rebuild it with the full evidence.
Read the old case study- My role
- UX Designer · interaction, visual, prototyping, dev
- Team
- Micah Watanabe (stakeholder) · Zeinab Serhan (PO) · Harshil Patel (dev)
- Timeline
- 6 months · May – Nov 2024
- For
- 3rd and 4th graders · teachers
- Learner engagement
- 60% → 85%
In 30 seconds
Legacy modules had unclear navigation, weak feedback, and passive videos kids skipped.
Lean on what engaged kids in pilots: stories and games. Every strategy became a planet with a game.
Engagement rose from 60% to 85%. The team also saw a 25% gain in reading comprehension.
Five reading strategies, taught by an AI tutor.
In plain wordsIt is a reading coach on a computer. Kids read a short text, try a strategy like "say it in your own words", and the software listens and gives feedback right away.
iSTART stands for Intelligent Student Tutoring for Active Reading and Thinking. Students practice five strategies with grade-level texts. Speech recognition, text-to-speech, and language processing give them feedback as they go, and teachers follow progress in their own view.
- Ask ItAsk questions about the text
- Reword ItSay it in your own words
- Find ItFind the key detail
- Explain ItExplain why it happens
- Summarize ItSum up the main idea
Built for high schoolers, used by 8-year-olds.
Unclear navigation, weak visual hierarchy, and almost no feedback.
Students dropped off mid-lesson or skipped through passive videos.
Younger kids needed fun, game-like experiences, not a high-school tool.
Understanding a student's world.
Immersed in prior research
Studied past iSTART and iSTART Early papers to understand the learning goals.
Synced with stakeholders
Regular check-ins with the product owner and stakeholder on expectations and limits.
Observed what engaged kids
In early pilots, storytelling and interactive games drove the most engagement.
Crewmate Quest: a paraphrasing game kids replay.
Inspired by Among Us. An alien has taken over your spaceship, and answering paraphrasing questions is how you take it back.
- Goal
- Make paraphrasing fun, engaging, and replayable
- Narrative
- An alien invades your ship. You, the crewmate, answer questions to take control back.
- Mechanics
- 2 chances per question · correct, you score · wrong, the alien scores
Five planets, five strategies, three stations each.
Practice became a journey. Every planet has the same three stops, so kids always know what comes next.
- Ask It
- Reword It
- Find It
- Explain It
- Summarize It
Students watched passive videos, and engagement stayed low.
Story-based interactive modules, where each screen builds curiosity.
Icons or words? We used both.
I designed icon-based buttons, which are easier for young kids to scan. The stakeholder preferred text only. We met in the middle.
Go deeper: the UX audit of 10+ legacy gamesWhat we fixed across the older games+
10+ legacy games reviewed
Found usability and accessibility issues across the old games.
Clearer and livelier
Rebuilt visuals, clarified instructions, improved animations.
Stronger contrast
Better color contrast, visual hierarchy, and game mechanics. WCAG 2.2 accessible.
Engagement went from 60% to 85%.
Learner engagement
Reading comprehension, a result shared with the learning science team
Accessible on desktop and mobile
What building for kids taught me.
- Kids learn better through stories, not just content.
- Stakeholder collaboration improves decisions, but it takes compromise.
- Game UX is never just visuals. It's feedback, loops, and motivation.
Allie
An AI shopping assistant people can steer.
Your move.
Open to Product Designer roles anywhere in the US.
gaurav.inbox29@gmail.com