Labs

Active builds. Production code shipping daily across regulated SaaS, viral B2C, AI governance tooling, and personal infrastructure. Engineering credibility you can verify — not slides, not roadmaps.

AI Agent Audit

AI Agent Governance & Reliability2025–present

Production audit daemon for AI agent installations in regulated environments. Always-on monitoring, periodic audit sweeps, configuration validation, auto-remediation, and reporting. Built for teams where errors compound silently and audit trails matter.

ThisWeekEats

Multi-platform SaaS2024–present

Multi-platform SaaS built solo using Claude-driven tooling. Backend (Node/Express), admin (Next.js), mobile (Flutter), worker process. Stripe-billed, AI-integrated meal planning. ~130k LOC of active code with regular release cadence. Live evidence of what one operator can ship with current AI tooling.

ReadReceipt

Viral B2C game · PLG2024–present

Daily game that exposes how well your friends actually know you. Lock 5 answers, send a link, watch them get it embarrassingly wrong. Viral mechanic built-in. Live PMF signal — daily-use.

The AI Reckoning

BookForthcoming 2026

Why most AI projects fail in production and what that's going to cost. 30 years of watching the same failures repeat across seven industries — and what current AI tooling is doing to that pattern.

Guardian

Personal AI memory system2025–present

Self-hosted persistent memory + decision support. Semantic memory with vector search, goal tracking, pattern discovery across decisions. Coordinates Claude, GPT, and Gemini based on task type. Production MCP server integration. Credibility artifact, not a SaaS.

PenForge

Voice modeling2025–present

Voice-calibration tool that scrubs AI-marketing-voice from generated drafts and matches a writer's actual cadence. Built because every default AI writing model reaches for the same shapes — predictable, generic, and easy to spot.

LearnMyVoice

Voice modeling pipeline2025–present

Pipeline that builds a voice model from a writer's existing corpus — interview transcripts, blog posts, written drafts. Powers PenForge and AI Reckoning drafts. Adjacent infrastructure for the broader voice-and-ghosting problem.

QTown

Autonomous AI experiment2024–2025

Autonomous-agent experiment — letting an LLM run a small simulated town with persistent state and emergent behavior. Case-study source for what current AI can and cannot do without supervision.

Want to talk about one of these in production? Get in touch.