Team
Macro Mixer
Project Concept
Generative UI for structured, scaled inference. Letting operators compose multi model research cascades with adjustable lenses, budget votes and receipts.
Entry
Status: Submitted
Last saved: May 09 at 5:07 PM HKT
Team Roster
You must be registered for the event to view the team message board.
Michael Ruescher Team Lead RSVP Approved
Founder at bitter.sh
Michael led the implementation and system architecture. He built the Next.js / React prototype, the card-based work surface, the expand and reduce API routes, the Gemini streaming integration through `@google/genai`, and the AG-UI event constructors using `@ag-ui/core`. He also implemented SSE parsing, board state management, card provenance (`parentIds`, directives, lens metadata, provider/model attribution), local session persistence, and the demo-ready map / reduce workflow.
0 to 1 product builder. Focused on long horizon interdisciplinary agents.
autonomous software development and business formation
bitter.sh
Matthias Heim RSVP Approved
Founder at Lailix
Matthias contributed technical and UX review of the board mechanics, including how generated cards should stream, how users should select and combine outputs, and how provenance should remain visible after reduction. He helped evaluate the implementation tradeoffs around keeping the prototype lean with Gemini streaming while preserving an event-driven architecture compatible with AG-UI and future A2UI-style projections.
Biologist turned AI builder. Spent a decade at Labster building virtual science labs used by millions of students, starting as a content creator and ending as AI Solution Specialist. Founded Lailix in October 2024 to help companies build practical AI systems, not slide decks. ETH Zurich background (MSc Biology), Swiss Study Foundation. Now based in Hong Kong, serving clients across Switzerland, Denmark, and Asia. We train product teams to build their own AI agents and ship working prototypes in days.
AI agent orchestration patterns and multi-agent systems. How non-technical domain experts can build their own AI tools. Voice AI for healthcare and compliance-heavy industries. The intersection of biology and AI (protein folding, genomics, drug discovery). Building sustainable AI consulting businesses. Swiss and Asian tech ecosystems, especially where they overlap.
Building AI agent systems that make knowledge workers 10x faster. Current projects: training Swiss enterprise teams to build their own AI agents (BlueCare, Brame), ad placement intelligence platform with real-time benchmarking, and VoiceDeck.ai for interactive voice-guided presentations. Also maintaining a personal "Brain" system where AI agents handle email triage, meeting prep, and CRM updates autonomously.
James Zheng RSVP Approved
student developer at -
James contributed product framing, demo narrative, and user workflow pressure-testing for the Macro Mixer board. He helped shape the map / reduce interaction around a real research workflow: expand a question into competing perspectives, compare the generated outputs, and reduce the useful subset into a synthesis. He also supported hackathon positioning around generative UI as an interactive work surface rather than a chatbot.
18 | Student developer & OSS contributor | improving accessibility with assistive technology
I learn by building and do my best solving problems and empowering others through tech and innovation.
Let's connect to build something together or simply chat about anything.
machine learning research, agentic AI, backend development, accessibility / assistive technology, and any random cool projects
Instagram CLI: Instagram's CLI and TUI client -- The ultimate weapon against brainrot
https://github.com/supreme-gg-gg/instagram-cli
- Stay connected with people while minimizing distractions
- Let your AI agent use Instagram