Life: Being a dad to Emilia. Making room for fitness and coffee along the way.
Building: This site, slowly and deliberately. Working closely with founders from the first questions onward, mostly between Tel Aviv and San Francisco. Putting the relationships and infrastructure around that work in place, and writing as my thinking develops.
Reading & Exploring: Genetic and evolutionary algorithms working with fleets of AI agents. I’m curious about what they could make possible in scientific research and engineering.
Thinking about:
Memory Systems: What will become the winning memory paradigm for the agentic era? How agents carry context across sessions, decide what deserves to persist, and revise what they think they know. Plenty of directions still open.
Security Infrastructure: What would let us trust agents with more responsibility? Identity, permissions, isolated execution, and a record of what happened across a session. I’m exploring this through small security prototypes and in Security Infrastructure Is Part of AI Infrastructure.
Architecting Agency: What does high-leverage knowledge work look like when cognitive load is the constraint, not time? My own experiment started as Soba and has evolved into Generalized OS: shared knowledge in Notion, agents in Grok Bot, and reusable skills and assets, from my design system to legal documents and workflows.
It reaches into fitness and coffee, too, with services for WhatsApp and xBloom. The aim is to carry what I’ve learned into the next task, with less context to reconstruct and more attention available for the person or problem in front of me.
Somewhere between San Francisco and Tel Aviv. Occasionally New York or Paris.