Muhtesem is an AI Engineer community speaker who worked with Re-skill and Diffusion Studio on open-source agentic video editing for personalized learning. At the 2025 AI Engineer Summit, he outlined a system that replaces dedicated rendering infrastructure with browser-native video rendering.
The Python-based agent uses Playwright to operate a specialized browser editor and generates JavaScript or TypeScript compositions with Diffusion Studio Core. Chromium renders the resulting videos through WebCodecs, while the Chrome DevTools Protocol transfers files between Python and the browser.
Its distinctive design combines:
- Code-generated video compositions: Agents express complex editing actions as executable code.
- Retrieval-grounded visual feedback: Documentation search and sampled video frames guide successive revisions before rendering.
- GPU-accelerated remote browsers: Separate browser sessions connected through WebSocket support concurrent editing agents.
- Agent-readable llms.txt documentation: Structured documentation helps agents locate implementation guidance.
A TypeScript implementation was planned when Muhtesem described the collaboration in his AI Engineer Summit presentation.
