


Featured · AI product
Velum
A private dream archive and AI-guided dream work app, designed and shipped end to end by one person.
- React Native
- Claude API
- Firebase
- Problem
- Dreams vanish within minutes of waking, and existing journals are just notes apps. There was no tool for capturing dreams quickly, seeing patterns across months of entries, or working with a dream in a structured way, all while keeping deeply personal data private.
- What I built
- A React Native iOS app with custom native modules, built using an AI-assisted development workflow with Claude Code.
- A privacy-first architecture: Firebase handles only auth and user claims, while all dream content is stored privately and never used as training data.
- A custom orchestration layer over the Claude API that enriches every entry by extracting symbols, emotions, people, places, and themes, and categorizes the archive automatically.
- Multi-step AI workflows for six structured dream work methods, including Amplification, Re-entry, and Gestalt Dialogue, run as guided, stateful sessions.
- Pattern-visualization “Boards” (recurring motifs, emotional patterns, frequency over time), a searchable archive, voice capture, and three visual themes.
- The full launch surface: marketing site and published privacy, AI, and data retention policies.
- Outcome
- In public TestFlight beta, ahead of App Store release.
