# Nanang

- **Event:** [Meta Global AI Developer Hackathon](https://cerebralvalley.ai/e/meta-ai-developer-hackathon)
- **When:** Oct 26, 12:00 PM – Nov 4, 9:00 PM (EST)
- **Where:** Online
- **Team:** [Nanang Mahdaen El Agung](https://cerebralvalley.ai/u/mahdaen)
- **GitHub:** https://github.com/beerush-id/wedyan
- **Demo video:** https://youtu.be/sTCegowGgx8
- **Gallery:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/1136

I make content on Facebook and I’m moving into short dramas. Making one means switching between a dozen apps for scripts, characters, pictures and editing, so I built the desktop I wanted: one place where I say what I want and an agent works with me, in the same apps.

Wedyan is a web-based desktop app you share with AI agents. You and the agent use the same apps, the same files and the same tools, and the agent works from the apps’ real state, not from screenshots.

Agents that operate software today look at pixels and guess. But every app already knows its own state: what is open, what is selected. Wedyan gives the agent those facts as data. Every change is one function, so the agent acts on what is real which a click calls and the agent calls too.

You ask by text or voice and agent answers on the desktop, it shows text, speaks, draws widgets, and asks with forms it composes from the desktop’s widgets. It acts in the apps at the moment it says so, and it keeps long tasks going by asking the system for the turns it needs.

Everything is a file in your home, stored on Cloudflare R2. Anything opens from a link, memory recalls by content, and the formats are open and documented.

The showcase is Motions app, a place to create video project including its script, its cast, and a stage per episode, with clips on a timeline. Its characters are designed in Lounge and look the same in every shot. Each stage has stills and takes for every shot, plus sound. You can edit a project by hand or by asking, and it renders to MP4 in the browser (rendering needs Chrome’s HTML-in-Canvas flag, preview works without it).

Everything in the repository was built during the hackathon. It's built on top of AirLib and AUI, my open-source libraries existed before the event.

Try it live: https://wedyan.app/?access_key=no9MtcEguZH9sI
PS: When the hackathon’s Baseten credits end, the router switches from Muse Spark Open to Muse Spark 1.3. Everything else works the same.

## More from Meta Global AI Developer Hackathon

- [PuzzleAR](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/1133)
- [Iqbal's Basilisk Lab](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/1134)
- [QuoteDesk](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/1135)
- [KineticLab](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/1137)
- [Esmada Zimba](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/1138)
- [JolSetu](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/1139)

---

Markdown version of https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/1136. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt