What We Build for Ourselves

AI-generated social content for home service businesses.

Multi-tenant .NET 10 platform that writes platform-specific posts by trade, season and channel. Built end to end, then paused so we could put the time into client work.

Household inventory that cuts food waste.

Browser barcode scanning, calculated shelf life across 500+ food categories, and recipe suggestions that use what expires first. Paused.

A hosted version of our code review pipeline.

The review pipeline we run on real enterprise changesets today, being rebuilt as a standalone product. The pipeline works now as a service. The hosted product does not exist yet.

One local panel over four AI coding CLIs.

Launches and restores sessions across Claude Code, Codex, Antigravity and Grok, shows what each one costs, and lists the tool calls that say they sent something off the machine. It runs on loopback only and never sits between us and the terminal.

One Playwright hub for every application we maintain.

Shared sign-in for forms and SAML, a scanner that crawls an application and maps its pages before tests are written, and a recorder that turns a click-through into page objects and test scripts. Before any test that changes data, a guard proves which environment the run is pointed at. Its dashboard, TestProof, reports what each run proved.

Mailing and multi-carrier shipping under postal and HIPAA rules.

One platform for mailing list processing and shipping: CASS and NCOA postal processing, real-time rate shopping across 10 carriers, and a security layer built for HIPAA-regulated mailings.

A junior golf team app for the phone and the web.

Coaches assign drills, players log rounds and practice, and parents follow along. One codebase runs on Android and in the browser, and scores recorded with no signal sync later without duplicates.