IteraShop² — in-house venture
A make-to-order microfactory that runs on its own agentic operating system
Our digital manufacturing venture integrates the chain from demand to the shop floor — CNC, 3D printing, and laser — on an operating system we built for ourselves. It makes real physical product to order: decorative lighting, branded stamps, cutters and rollers, automotive and marine parts, and industrial fixtures. It is the proving ground where our stack earns its way into client work.
IteraShop² is a digital microfactory built around make-to-order: AI integrates the chain from demand to a CNC, 3D-printing, and laser shop floor, so a piece is produced when it is ordered rather than held in inventory. It is also deliberately the laboratory of IteraSystems — every pattern in our stack runs here before it ever reaches a client.
The output is real physical product. LUVO Lamps, our lighting brand, is the most visible line; alongside it, the shop makes branded stamps, cookie cutters, and embossing rollers for our own brands and for client businesses such as bakeries, plus automotive parts, marine components, and fixtures for industrial processes. Short-run, made-to-order work is exactly where a digital microfactory beats a warehouse.
The venture is organized around an explicit value chain — three engines (market, product, execution) spanning eleven processes from market intelligence to fulfillment and learning — and the operating system instruments that chain rather than a generic ERP template. Costing is parametric per piece — energy, labor, maintenance, rework — feeding a deterministic quote engine.
Production runs, incidents, and operator checklists are tracked in the same system, deployed at the edge. The catalog and marketing pipeline is operated with AI agents, and a small team coordinates through GitHub-based workflows. When we tell a client their operation can run on an agentic system, this is the existence proof.
The storefront you can visit runs on this pipeline
Every piece travels the same route: an operator-owned master workbook, normalization into a frozen schema, a human legal gate that decides what may be sold, and one JSON artifact the shop renders. The counts below are the real ones, read from the catalog package when this page was built.
Operator-owned sheet · single source of truth for every piece.
Cleaned into a frozen schema · 22 of 23 carry a photo.
21 held in evaluation · human sign-off, publishable status only.
Regenerated by script, never edited by hand · versioned with the code.
21 shown as in evaluation — visible to browse, not for sale.
Counts on this sheet are imported from @itera/catalog at build time — the same data the storefront renders. 23 and 2 here mean 23 and 2 there.
Evaluating your first serious AI implementation? Let's look at your operation together.
Tell us your company and the process that hurts most. The diagnostic is paid, scoped in writing, and graded on evidence.
Request a diagnostic- linkedin.com/in/arturomunozcruz
- GitHub
- github.com/arturo-a11y