3x founder, 2 exits via acquisition, 3 US patents in AI systems. I build agentic systems where reliability and UX matter as much as the model—from scaling ProAI through acquisition to running autonomous agents on edge hardware. Currently focused on agent-first product design: why ad-hoc route planning beats rigid graph workflows, and what it takes to ship AI that actually works for users.
WRITING
A framework for agents that evaluate, modify, and evolve their own behavior—including the self-improvement.skill library.
Why the guardrails we built to make agents safe are now the things making them dumb.
The Agentic Shift: 2026 Playbook Jan 2026
A quantitative analysis of why 95% of AI pilots fail.
Why true agentic efficiency requires abandoning human workflows.
PROJECTS
- US Patent PendingSystem that generates competitive strategy documents from market data using multi-step LLM reasoning.
- US Patent PendingAutomated market research utilizing a plurality of linked AI agents for comprehensive industry analysis.
- Open Source / GitHubVisual builder for LangGraph agents. Drag-and-drop agent design without losing access to the underlying graph.