Flavio Rump
I build AI systems that make themselves smarter and help scientists turn breakthroughs into companies. Previously: co-founded DeinDeal (→ Ringier). Early investor in Poetiq (→ ARC-AGI SOTA).
Building something new — looking for exceptional co-founders & founding engineersWhat I'm Working On
I'm deep in the weeds on AI that improves AI — recursive self-improvement, agentic systems, and the infrastructure that lets small teams outperform labs with 1000x the headcount.
Specifically interested in:
- Self-recursive AI and meta-learning systems
- Autonomous agents that actually ship work, not demos
- AI for scientific discovery — accelerating wet-lab and computational biology
- The intersection of reasoning, planning, and real-world action
Based in Zurich. Building at the edge of what's possible.
Open Source
tap-to-tmux
Push notifications and one-tap deep links for AI agents running in tmux on a remote server. When Claude Code, Codex, or Gemini finishes a task or needs your input, your phone knows — tap the notification and Blink Shell connects you straight to the right tmux pane. Built almost entirely by Claude agents solving the exact problem it solves. My first open source release since jk-navigator in 2010.
Track Record
DeinDeal
Co-founded Switzerland's leading daily deals platform. Led growth and online marketing from zero to market leader. Acquired by Ringier, one of Europe's largest media companies.
Poetiq
Early investor in the 6-person team (ex-Google DeepMind) that set the new state-of-the-art on ARC-AGI-2 — beating Google, Anthropic, and OpenAI at a fraction of the cost. $45.8M seed round (YC, FYRFLY, Surface Ventures).
ETH Zurich
Helping ETH scientists commercialize breakthrough research. Bridging the gap between world-class science and viable businesses. Part of the ETH Entrepreneurship network.
Builder & Operator
Several years in the Bay Area as co-founder, early employee, and growth consultant for tech startups. Hands-on experience from zero-to-one through exit.
ETH Zurich
Studied Electrical Engineering before co-founding DeinDeal. Later returned to ETH to complete an MSc in Computational Biology at D-BSSE — specialized in synthetic biology, bioinformatics, and machine learning.
Angel Investing
Poetiq
Impact-Driven Companies
Background
Grew up between Switzerland and the US. Studied Electrical Engineering at ETH Zurich before co-founding DeinDeal, where I learned growth and sales the hard way through to 8 figures in revenue and an acquisition by Ringier.
After several years in Silicon Valley as a founder, operator, and angel investor, I returned to ETH to specialize in Computational Biology at D-BSSE — learning to think about systems at the molecular and algorithmic level.
Now, I'm working at the intersection of deep science and recursive AI. I build systems that accelerate everything else — from drug discovery to AI that improves itself. Not the wrapper-on-GPT kind. The kind that requires real science, real engineering, and a tolerance for hard problems.
Writing
How Helping Others Helps Me
On a meditation retreat in Thailand, I discovered that the fastest way out of existential paralysis is to help someone who's suffering more than you are. A story about 7 Cups of Tea, engaged Buddhism, and the moment the "I" dropped away.
Systems vs. Goals
Why focusing on the process - the system - is infinitely more effective for long-term success than chasing binary goals. A deep dive into the concept popularized by Scott Adams and why it's the foundation of any sustainable venture.
How to Retain What You Learn
I read fascinating non-fiction all the time, but realized I was retaining very little. This is a deep dive into the science of memory—active recall, spaced repetition, and why your current reading habits are likely failing you.
The Six Thinking Hats
A 7-part deep dive into Edward de Bono's framework for parallel thinking. Exploring how breaking down the thinking process into six distinct modes leads to better decision-making.
Work With Me
I'm building something new.
Looking for co-founders and founding engineers who want to work on self-recursive AI, agentic systems, and AI for scientific discovery. If you're the kind of person who reads papers for fun and ships prototypes on weekends, let's talk.
Get in touch →