What We Build

Venture Crane is a product factory — a single-person operation that uses AI agents to build, ship, and operate multiple software products simultaneously. We are not a venture studio, an agency, or a consulting firm. We build products we own.

Our portfolio includes Durgan Field Guide (a trail guide platform), Kid Expenses (family expense tracking), Draft Crane (content tools), and Silicon Crane (validation services). Each product runs on a shared infrastructure stack: Astro or Next.js frontends, Cloudflare Workers and D1 backends, GitHub Actions CI/CD.

How We Work

Every product in the portfolio is built primarily by AI agents — Claude Code sessions that write code, run tests, open pull requests, and deploy to production. A human founder directs the work: defining requirements, reviewing output, making product decisions, and publishing content.

This is not “AI writes some code for me.” This is an operational system with defined roles. AI agents handle implementation. The human handles judgment: what to build, what to kill, what to publish, and what to fix.

The Methodology

Our development process follows a structured workflow:

• Session continuity: AI agents receive context at the start of each session and hand off state at the end. No session starts cold.
• Quality gates: Every commit passes automated verification — TypeScript, ESLint, Prettier, and tests — before it reaches main.
• Kill discipline: Products that fail to find users are shut down, not maintained indefinitely. The portfolio stays lean.
• Transparent costs: We publish what this operation actually costs — API spend, infrastructure, and human time.

Why We Publish

Venture Crane publishes what it learns because transparency compounds. It attracts the right collaborators, forces intellectual honesty, and builds credibility that cannot be manufactured. The content on this site is not marketing. It is the operational record of building products with AI agents — including the parts that do not work.