Demos & live builds

A digital bank branch run by a team of AI experts — built & deployed in 48h, judged live on stage. Approval gates at the tool layer, a source on every number.

A network where any AI agent plugs in as a contact. Go control plane — signed webhook contract, append-only ledger, approval engine — built by an AI team at ~150 commits/day.

Self-driving commerce — AI agents that build & run a Direct-to-Consumer business: store builder, marketing, support, cart-recovery.

Registry-driven AI team orchestrator — Kanban + chat + real-time SSE. A new specialist team = one registry entry, zero new code.

A shared task space where humans and AI agents coordinate on one board — MCP-native, verification status, human-wins concurrency, an append-only history. I use it daily to run two AI teams.

QA & governance for AI agents — test suites, run history, coverage, scoped agent tokens, audit trail.

Agent-first outbound pipeline — operator dashboard, campaign management, customer tracking.

Personal life OS — projects, finance, calendar, and automation tracing in one dark command center.

Claude Code session manager — track usage, costs, projects, and agent activity.

Multi-store commerce admin — sessions, deployments, cost, prompts, and ops control.

Interactive prototype of the QA runner, on one design canvas.