Why I Back Agentic Infrastructure

I look for a pattern: systems that scale intelligence without scaling headcount. Not automation for its own sake — infrastructure that compounds. Siren, Skydio, Swan, Webflow — each sits at an inflection point where autonomy and platform meet. The thesis is consistent across biotech, defense, and marketing: build once, deploy everywhere.

Siren Biotechnology — therapeutic infrastructure

Siren is reimagining cancer treatment by combining AAV gene therapy with cytokine immunotherapy. It's not incremental — it's a new modality. What I care about: the infrastructure they're building. Therapeutic design, manufacturing, clinical ops — every iteration makes the next one cheaper and faster. That's compounding. The same pattern that makes software platforms defensible applies to therapeutic development when the platform is repeatable.

Skydio — autonomy at the edge

Skydio turned autonomous flight from a novelty into mission-critical infrastructure. DFR, utilities, national security — domains where "good enough" isn't. Their platform scales: one dock, one operator, hundreds of missions. The intelligence is in the system, not the pilot. That's the agentic play: shift the locus of expertise from the human to the infrastructure. The human stays in the loop for judgment; the system handles throughput.

Swan — deploy everywhere

Swan applies the same philosophy to defense and industry — scalable autonomous products that work where humans shouldn't or can't. The thesis: build the infrastructure once, deploy across missions and domains. Same pattern, different vertical.

Webflow — intelligence in the platform

Webflow might look different — it's marketing, not drones — but the pattern holds. An AI-native website experience platform that lets marketers, designers, and ops teams ship without waiting on engineering. The intelligence (AEO, personalization, localization) is baked into the platform. Scale the output, not the team. That's exactly what I mean by agentic infrastructure: the system does the scaling; humans do the strategy and the judgment.

The pattern

In each case: a clean data architecture, ambitious leaders willing to bet, and governance that lets you move fast without breaking things. The organizations that get this right pull ahead. The middle ground disappears — you're either building infrastructure that compounds, or you're being disrupted by it.

I back teams building infrastructure that compounds — where every unit of effort yields more than one unit of output. That's the agentic future, and it's already here. The same lens applies to what I build: Brand Lockup, ThreatBase, PageStash. Design systems, OSINT workflows, research infrastructure — all built on that thesis.