# Alkahwaji

- **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:** [Abdelrahman Alkahwaji](https://cerebralvalley.ai/u/Alkahwaji6)
- **GitHub:** https://github.com/Alkahwaji/scientific-discovery-studio
- **Demo video:** https://youtu.be/korrelbqoQk
- **Gallery:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4429

# Scientific Discovery Studio
### Creative Thinking Techniques-Based Multi-Agent AI for Scientific Discovery

**Project Description**

Scientific Discovery Studio is an agentic AI platform designed to transform scientific research questions into creative, evidence-supported, and experimentally testable hypotheses.

Unlike conventional AI research assistants that primarily rely on large language models and general-purpose prompting to generate ideas, Scientific Discovery Studio introduces structured creative-thinking methodologies directly into the scientific discovery process.

Powered by Meta Muse Spark 1.3, the platform orchestrates specialized AI agents implementing Brainstorming Varieties, Morphological Analysis, TRIZ, and SCAMPER to explore research questions from different perspectives and generate diverse mechanistic hypotheses.

The platform integrates a comprehensive research pipeline encompassing multi-source literature retrieval (Europe PMC, OpenAlex, PubMed, and arXiv), scientific paper analysis, targeted prior-art verification, and causal mechanism validation.

Its iterative evaluation engine combines scientific scoring, comparative ranking, adversarial debate, hypothesis evolution, reflection, and meta-review to progressively refine candidate hypotheses.

For computational validation, Muse Spark plans simulations while Muse Code implements and executes Python models within its sandboxed environment. Finally, an Experiment Design agent translates promising hypotheses into proposed validation experiments.

Researchers can interact with the platform through text or Muse Voice Transcribe and customize research cycles, API usage, literature sources, and hypothesis-generation limits.

**The outcome is a structured, reproducible research workflow producing ranked hypotheses, literature evidence, causal critiques, simulation results, and experimental proposals.**

Our central philosophy is simple:

**We don't just ask AI to be creative. We give it structured methods to think creatively.**

Scientific Discovery Studio aims to make AI a more capable research partner, helping scientists explore overlooked possibilities, challenge existing assumptions, and accelerate the journey from research questions to testable scientific hypotheses.

## More from Meta Global AI Developer Hackathon

- [CatchLength](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4426)
- [Meeting Copilot](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4427)
- [Glint – Parvathi Nalla](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4428)
- [Muselancer](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4430)
- [rehearsal](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4431)
- [Muse Inkline](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4432)

---

Markdown version of https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4429. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt