Literally

Inspiration

Literally was inspired by a growing decline in sustained attention caused by short-form digital consumption. Modern feeds optimize for rapid engagement rather than comprehension, resulting in fragmented focus and cognitive fatigue. At the same time, research in neuroplasticity shows that attention and reading capacity remain trainable even in late adolescence and adulthood. We saw an opportunity to redesign digital consumption itself — not by eliminating technology, but by rebuilding it as a system that strengthens attention rather than eroding it. From the beginning, we intentionally designed Literally to combine adaptive AI, privacy-first infrastructure, multimodal generation, and intelligent curation so that it could meaningfully impact education, safety, and digital well-being at scale.

What it does

Literally is an AI-powered Android platform that transforms real-world content into adaptive cognitive training. Users select topics in fast-moving domains such as AI, crypto, geopolitics, and technology. The system curates up-to-date articles, generates AI-adaptive summaries that increase in complexity over time, and reinforces comprehension through gamified quizzes and mini-games. Screen-time analytics and a visual data flow map provide transparency, while anonymous discussion threads encourage thoughtful long-form discourse. Child accounts automatically rewrite unsafe language and provide parent-facing summaries. The result is a structured progression engine that converts passive scrolling into measurable attention development while maintaining privacy and trust. The app reinforces comprehension through gamified quizzes and mini-games that reward depth over speed. It also provides structured article stacks, screen-time analytics dashboards, and a visual data flow map to ensure transparency around usage and data handling. The result is a guided progression system that turns passive scrolling into measurable attention training.

How we built it

We architected Literally as a modular AI ecosystem integrating all hackathon partner technologies. AWS Bedrock AgentCore powers our backend orchestration layer, managing content ingestion, adaptive summarization agents, moderation workflows, and progression logic. MiniMax APIs are used for multimodal generation, including AI voice summaries, animated explainers, interactive quiz narration, and safe-content rewriting for child accounts. OAX-inspired intelligent curation logic ensures that content in fast-evolving domains is timely, credible, and relevant. RevisionDojo-style adaptive mastery principles guide our progression engine, which dynamically adjusts reading complexity based on behavioral performance. Abelian/QDay quantum-resistant infrastructure secures user data with encrypted, privacy-preserving design, enabling anonymous community participation and safe personalization. ExpressVPN-aligned digital guardian principles shape our privacy-first scraping, secure data transport, and child-protection systems. Each technology is embedded into core functionality, qualifying Literally across all award tracks through genuine technical integration rather than superficial inclusion.

Challenges we ran into

A major challenge was balancing engagement with discipline. Many digital platforms rely on overstimulation to retain users, but our goal required sustained focus and progressive cognitive load. Designing gamification that rewards comprehension rather than impulsivity required iterative behavioral tuning. Integrating multiple technologies across AWS agentic orchestration, MiniMax multimodal generation, and Abelian privacy infrastructure introduced architectural complexity, particularly in maintaining secure data flow while enabling adaptive personalization. Ensuring ethical content curation in fast-moving domains also required credibility filtering and dynamic ranking logic. Finally, building child-safe rewriting systems without resorting to invasive surveillance demanded careful privacy-first engineering.

Accomplishments that we're proud of

We successfully built a working adaptive reading system that dynamically evolves with user performance and integrates AWS agentic AI, MiniMax multimodal capabilities, Abelian privacy infrastructure, and OAX-style intelligent curation into a unified platform. We are particularly proud that Literally qualifies for every hackathon award track through authentic technical implementation: agentic AI via AWS Bedrock, creative multimodal usage via MiniMax, privacy-preserving architecture via Abelian/QDay, children’s safety alignment with ExpressVPN principles, adaptive education aligned with RevisionDojo, fast-domain EdTech impact aligned with OAX, and entrepreneurial scalability aligned with HKUST EC. Rather than treating each sponsor as a separate add-on, we designed Literally as a cohesive AI infrastructure platform.

What we learned

We learned that meaningful AI systems require both behavioral design and architectural rigor. Adaptive learning is most effective when grounded in measurable progression signals, and privacy must be embedded at the infrastructure level rather than added later. Integrating multiple AI services demands clear orchestration logic and modular design. We also learned that transparency significantly increases trust; users respond positively when they understand how their data flows and how AI decisions are made. Most importantly, we learned that technology can either fragment attention or strengthen it depending on how it is designed.

What's next for Literally

Next, we plan to refine our adaptive progression model using expanded behavioral datasets to optimize cognitive growth pacing. We aim to expand multimodal features further using MiniMax APIs, improve AWS agent coordination for real-time personalization, and deepen Abelian-based privacy integrations for user-owned encrypted cognitive profiles. We also plan to strengthen institutional partnerships in education and corporate productivity sectors, positioning Literally as a scalable AI-powered attention training platform. Long term, our goal is to formalize attention growth metrics and evolve Literally into a privacy-preserving, agent-driven cognitive infrastructure platform that continues to qualify across education, AI, privacy, and innovation domains beyond the hackathon.

Built With

• amazon-web-services
• expo.io
• minimax
• reactnative
• xml