Chronorama

Inspiration

Young people may be more naturally drawn to Minecraft than to a traditional history lesson, but the two do not have to be opposites. Games can increase attention and motivation, and a Minecraft-like sense of exploration can become a powerful way to learn.

Chronorama began with a simple question: what if learning history felt less like memorizing a lesson and more like stepping inside a small living world?

I had previously worked on a VR application for students learning French as a foreign language. It recreated everyday situations so learners could practise language in context. That experience showed me how effective digital environments can be when they let learners experience a subject rather than only read about it.

LLMs add another layer to that idea. Every student is curious about different things. One may want to understand tools and irrigation, while another is more interested in social organization, food, housing, or everyday life.

An open dialogue system lets learners follow their own curiosity instead of receiving exactly the same explanation as every classmate.

Chronorama can be imagined as a modern, playable pop-up history book: more immersive and personal than a textbook or chatbot, but more focused, accessible, and scripted than Minecraft.

What it does

Chronorama is an educational 3D web experience in which young people learn about human history by exploring interactive dioramas.

Instead of mainly reading fact sheets or answering quizzes, players move through a historical scene, observe its inhabitants, manipulate period objects, solve small contextual problems, and ask characters their own questions through an LLM-powered dialogue system.

The long-term vision presents human history as a timeline of miniature 3D worlds, including the Neolithic period, Ancient Egypt, Greece, Rome, the Middle Ages, the Renaissance, the Industrial Revolution, and the modern era.

The current hackathon prototype contains two playable dioramas from the beginning of that journey.

The first villages — Neolithic period

The player explores a small settled community, speaks with Alya, collects wood, plant fibres and flint, crafts a simple tool, repairs a shelter before a storm, and tends the village hearth.

The quest introduces ideas such as settlement, cooperation, agriculture, domestication, toolmaking, food storage, and the importance of maintaining fire.

Along the Nile — Ancient Egypt

The player meets Hori, a farmer whose field is no longer receiving enough water. They inspect and clear an irrigation channel, use a shaduf to lift water, carry measured grain, and report the result to Merit, a scribe.

The scene introduces the role of the Nile, irrigation, agriculture, mud-brick construction, grain storage, measurement, writing, and administration.

Across both scenes, the player can:

• move through and observe the environment;
• speak with historical characters;
• ask suggested or completely free-form questions;
• collect and manipulate objects;
• solve short contextual puzzles;
• use simple crafting and repair mechanics;
• unlock discoveries in a historical field journal.

How we built it

Chronorama was designed and built as a solo hackathon prototype in approximately two to three hours.

My role covered:

• concept and product direction;
• game and learning-experience design;
• UX design;
• visual direction;
• playtesting and iteration.

The implementation was produced through AI-assisted development with OpenAI Codex Work and GPT-5.6 Sol.

The playable application uses:

• OpenAI Codex Work;
• GPT-5.6 Sol for AI-assisted development;
• OpenAI Responses API for in-world character dialogue;
• TypeScript;
• React;
• Three.js;
• React Three Fiber;
• Drei;
• Vite and Vinext;
• HTML and CSS interface overlays;
• local fallback dialogue.

Chronorama does not present the LLM as a separate generic chatbot. The player speaks directly to a character who belongs to the historical world.

Each dialogue request can include contextual information such as:

• the historical period;
• the character's identity and occupation;
• the current location and quest;
• the player's progress and inventory;
• nearby objects;
• the selected explanation level;
• a curated historical knowledge base for that period.

Three suggested questions help young players understand what they are allowed to ask, while the free-text field lets them pursue their own interests.

Asking questions is also rewarded through additional journal discoveries, making curiosity part of the game loop rather than an optional chatbot feature.

If no OpenAI API key is available, Chronorama uses local predefined answers so the core experience remains playable.

Historical grounding and safety

The goal is not to let the model freely improvise history.

Each period should have its own curated knowledge base, built from relevant school curricula and carefully selected historical references.

Characters are instructed to:

• remain within their historical role and period;
• avoid knowledge of future technologies;
• answer briefly and at an age-appropriate level;
• acknowledge when something is unknown or historically debated;
• avoid inventing facts outside the supplied material;
• avoid implying that everyone in a long historical period lived in exactly the same way.

This structure is intended to combine the freedom of an LLM conversation with the reliability required for an educational experience.

Challenges we ran into

The prototype quickly demonstrated that producing a 3D scene is only the beginning.

The main challenge is polishing the experience:

• making the 3D scenes richer while preserving mobile performance;
• making every interactive object obvious and easy to select;
• separating camera gestures from character movement on touchscreens;
• keeping the interface readable without covering the diorama;
• refining educational quests so that learning emerges naturally through play;
• improving the LLM experience while keeping answers historically grounded.

Several iterations were necessary to improve mobile camera controls, enlarge interaction targets, simplify the interface, clarify quest steps, and prevent players from becoming blocked.

The 3D rendering, interactions, and mobile UI still need additional refinement, but the prototype already demonstrates the complete concept.

Accomplishments that we're proud of

I am most proud of the central concept.

Even in a very short prototype, Chronorama demonstrates how exploration, game mechanics, historical context, and open-ended AI dialogue can belong to the same learning experience.

The two scenes also show that the idea can extend across very different periods while reusing the same basic structure:

1. enter a historical world;
2. meet one of its inhabitants;
3. understand a concrete problem;
4. manipulate the technologies of that period;
5. ask personal questions;
6. leave with contextual historical knowledge.

The prototype was designed and made playable in only two to three hours, which also demonstrates how quickly AI-assisted development can turn an educational concept into something that can be tested with real users.

What we learned

AI coding tools are changing the bottleneck of prototyping.

The hardest part is increasingly not writing the code itself. It is choosing a meaningful pain point, designing an experience that genuinely responds to it, and putting that experience in front of potential users.

Chronorama was produced extremely quickly, but its real value will only become clear through playtesting:

• Do young players understand the interactions?
• Do they become curious?
• Do they remember what they learned?
• Do they voluntarily ask questions that a traditional lesson might never have prompted?
• Does the experience make them want to explore another historical period?

That is the next and most important test.

What's next for Chronorama

Before adding more historical periods, I would first refine the two existing dioramas:

1. polish the core exploration and quest loop;
2. improve touch interactions, UI clarity, and 3D fidelity;
3. deepen the contextual LLM dialogue system;
4. strengthen and review the historical knowledge bases;
5. playtest the experience with young learners and educators.

Only after the core experience feels convincing would I expand the timeline into Greece, Rome, the Middle Ages, the Renaissance, the Industrial Revolution, and later periods.

For now, Chronorama is a hackathon prototype and a question offered to the judges:

Can a small playable world make a learner want to explore our shared human history?

Built With

• codex
• gpt5.6
• three.js