Interactive educational videos that adapt to every choice.

TMochiLearn turns a single topic into an interactive lesson with choices, alternate explanations, and outcomes shaped by the learner. Describe what you want to teach, choose how deeply it should branch, and generate a complete learning-path tree.

Every path belongs to one connected learning experience. Learners can explore different approaches, see the consequences of each choice, and build a deeper understanding of educational and technical subjects.

Built with Codex using GPT 5.6 Sol in High and Ultra settings, with manual QA, verification, and code debugging.
Submission for the OpenAI Devpost hackathon.

Don’t just watch. Explore.

A TMochiLearn lesson presents clear, focused video segments and lets the learner choose what to examine next.

• A subtle Choose the next path cue appears before each decision.
• Learning-path previews transition naturally from the current topic.
• Every choice continues from the same moment without losing sound or playback settings.
• Multi-level lessons remember the path already taken.
• Shared lesson links open directly in the interactive player.
• The public learning library makes published lessons easy to discover and replay.

One topic. Every learning path.

The Creator Studio is designed for educators, trainers, and technical creators.

1. Describe your lesson — start with a concept, process, skill, or scenario learners should understand.
2. Choose the depth — create a focused fork or a learning tree with up to three levels of decisions.
3. Build the lesson — follow live generation progress and see the learning-path map fill with scenes.
4. Preview every route — explore random paths or review the lesson choice by choice.
5. Describe the lesson — write the public details yourself or let TMochiLearn generate a title and description.
6. Publish the experience — release it to the TMochiLearn learning library or download the complete artifact package.

Made for interactive learning

• Prompt-to-interactive-lesson creation powered by Samsar
• One-to-three-level learning trees with distinct branches and outcomes
• Live branch topology that shows the structure as it is created
• Resumable sessions so a lesson remains available at its own Creator URL
• Interactive previews with media-accurate decision points
• Smart publishing metadata generated from the completed lesson
• Transparent credit estimates and balances throughout creation
• Downloadable production artifacts for every rendered path
• A public learning library with search, featured lessons, and shareable watch links

Learning with more than one path

TMochiLearn is built around a simple belief: learners understand more when they actively explore. They can test an assumption, compare an approach, troubleshoot a system, or revisit a concept—and the lesson should respond to every decision.

The result is an adaptive format for educational and technical content: structured enough to teach intentionally and flexible enough to support curiosity.

Powered by Samsar

TMochiLearn uses Samsar to create, render, resume, and publish branching lessons. Existing Samsar accounts work in the Creator Studio, and completed interactive publications flow into the same public catalog used by learners.

For contributors

The project requires Node.js >=22.13.0.

npm install
npm run dev

Run the same checks shown by the badge:

npm run lint
npm test

npm test builds the Cloudflare Worker version and runs the existing Node.js integration suite against the rendered learning experience, shared player routes, Creator authentication, branching contracts, generation flow, and publishing tools.

Configuration and deployment notes

The Samsar API defaults to https://api.samsar.one/v1. Set SAMSAR_API_BASE_URL to use another API origin. Creator requests use the signed-in user’s shared Samsar bearer session. Creator image and video options are loaded server-side from the public /video/supported_models Express model catalog and submissions are revalidated against that catalog. Additional artifact hosts can be provided with the comma-separated SAMSAR_ARTIFACT_HOSTS value.

To preview the local UI against the live production publication catalog, run:

npm run dev:production-catalog

This only reads the public production catalog for learners. Creator actions still require a signed-in Samsar session.

For Vercel, use the Next.js framework preset and the standard npm run build output. The optional Cloudflare Worker workflow is available through npm run dev:worker, npm run build:worker, and npm run start:worker.

Demo

App demo screencast created with Codex using GPT 5.6 Ultra settings.