§ 03 · Writing
Notes on what's next, and how we decide.
Three formats, one fund. Essays work out a belief at length. Research reports show our map of a market. Field reports log what we've seen from the founders' desks we visit. Everything here is a prior — subject to revision.
Essay
A thesis, argued.
Long-form, literary. One idea, worked until it holds up. Written for the founder who wants to know what we actually believe.
Research
A market, mapped.
Structured notes on a sector — what's commoditizing, where margin is moving, who's building in the gap. Data, sourced; opinions, labeled.
Field
A founder, up close.
From the desk of the partner who wrote the check. What we saw, what surprised us, what changed our prior. Dated and signed.
Recent · 12 entries
- Mar 2026 essay The runtime is the moat. →
- Feb 2026 research Commoditization of the base layer, Q1 2026. →
- Feb 2026 field From the desk — on ticket-rs, a tool for coding agents. →
- Jan 2026 essay On priors: how we decide what to believe, early. →
- Jan 2026 research Mapping the agent-memory stack. →
- Dec 2025 field From the desk — sgrep, and the return of the static binary. →
- Dec 2025 essay Evals are the new dashboards. →
- Nov 2025 research A taxonomy of agent failure modes. →
- Nov 2025 field From the desk — SQLGenie, and the problem with schemas. →
- Oct 2025 essay Why agentic payments need a different rail. →
- Oct 2025 research Context windows as a capital allocation problem. →
- Sep 2025 field From the desk — a week at three retrieval-heavy startups. →
The prior, in your inbox.
One long piece a month, plus two short field reports. No product updates, no announcements.