Start a Project
proof of work

Case studies.

Not screenshots — systems. Each study walks through the problem we were handed, what we designed and built, and what actually changed once it was running. Everything here runs on infrastructure we own and operate.

the work

Problem → system → outcome.

A running list of builds we can talk about in detail — click any study for the full story. More are in flight; private pilots get published here once they ship.

CS·13

Equestrian event entry & live scoring platform

Online entries, automatic conflict-aware ride times, live scoring, and an AI concierge — a horse show run from one screen.

private pilotweb platformautomationlive scoring
Read the study
CS·12

Launch OS — from first lead to launch day

A lead-to-launch platform that carries every prospect through one pipeline — from automated lead sweep to a launch checklist that won't let a project ship half-finished.

in productionweb platformpipelineinternal tool
Read the study
CS·11

voxelbox.org — the studio site as a system

Our own site: hand-built, static-fast, a WebGL voxel hero — and a homepage that updates itself when new work ships.

livewebWebGLstatic-fast
Read the study
CS·10

Voxelburg — a living AI town

A persistent virtual town where AI residents live out their days — chatting, forming opinions, and remembering you between visits.

liveAI agentsself-hosted LLMssimulation
Read the study
CS·09

The edge — public products, zero open doors

Every public service we run reaches the internet through secure outbound tunnels — no inbound ports, origins that refuse direct connections, and private tools behind an access gate.

in productionnetworkingsecuritytunnels
Read the study
CS·08

The fleet — infrastructure that watches itself

A fleet of self-hosted agents that patrol the stack on timers, investigate anomalies, and report to Discord — often with the diagnosis attached.

in productionAI agentsmonitoringchat ops
Read the study
CS·07

Mission control — the whole stack on one screen

A self-hosted hub that answers 'is everything up, and what changed?' in one view — status, operational feeds, and common actions for every service we run.

in productiondashboardobservabilityinternal tool
Read the study
CS·06

The ops assistant — an AI teammate in Discord

An AI assistant that lives where the team already is — reading playbooks, running real checks, and answering with actions instead of links.

in productionAI assistantchat opsplaybooks
Read the study
CS·05

Client Radar — lead generation on autopilot

An automated sweep that finds local businesses with no web presence, enriches each lead, and drafts outreach — with a human approving every send.

in productionautomationdata pipelineinternal tool
Read the study
CS·04

The AI stack — models on metal we own

Language models, image generation, speech, and private search — served from our own AI hardware, with cloud models pulled in only when a task earns it.

in productionself-hosted LLMsimage genspeechinfrastructure
Read the study
CS·03

The demo pipeline — a live site in minutes

Idea to live, shareable demo in minutes — every mockup gets its own voxelbox.org subdomain, TLS included, no server work per site.

liveinfrastructurehostingweb
Read the study
CS·02

Custom 3D prints — a print shop run like a server farm

Made-to-order 3D prints — from idea or file to finished part — with the same queue, quality checks, and status comms we give servers.

live3D printingfabricationmade-to-order
Read the study
CS·01

Community game servers — where it all started

Free, well-run game worlds on hardware we own — the build that started VoxelBox, and the ops discipline everything else inherited.

livegame serversself-hostedcommunity
Read the study
your project

The next study could be yours.

Bring us the problem — the messy, manual, “there has to be a better way” kind. We'll design the system, build it, and run it.