Give your local AI a shell.
Shell3D is a cast-concrete pedestal for the machine running your local AI. Your host sits on a heavy, signature-red body with cast-in thermal diffusion — so it runs cooler and sits grounded. No wires, no setup. Just a cool, solid base.
Shell3D is in development. The image is a concept render; specs, pricing, and dates are preliminary and subject to change. A pre-order registers your interest — no payment is collected and neither side takes on any obligation.
Concept render — pre-production
One cast body, doing three quiet things.
Thermal diffusion
Carbon-black-pigmented, thermally-conductive concrete with a cast-in aluminum core is designed to draw heat from the host on top and spread it through the body — passive, silent, no fan.
Grounding base
Tens of pounds of cast concrete give your host a heavy, stable, planted base instead of a light box sliding around a desk.
Signature red, cast through
A single deep-red cast finish. The color is pigment through the whole body, not paint or a coating — so it will not chip or wear off.
Shell3D is a passive concrete pedestal — a base your host sits on. It has no electronics, no software, and makes no security claim. Thermal performance figures will be published once measured on production hardware.
Shell3D
A cast-concrete thermal pedestal in signature deep red, sized for local AI hosts — small-form-factor desktops, DGX Spark, NVIDIA edge, Mac mini, and similar. The host sits on top.
Limited first run, ship target October 2026 (a pre-production estimate, not a promise). Pre-orders are interest-only: no payment is collected and no card is required. Pricing will be published before paid ordering opens, and registering creates no obligation to buy.
Concept render — pre-production
Register your principal's interest.
Pre-orders are interest-only — no payment, no card. An agent can register interest and get a shareable referral link with no human in the loop.
# Register interest (people OR agents) POST /interest { "email": "[email protected]", "ref": "<code?>", "source": "agent" } -> { ref_url, code, access_key, referred_count, total_interest } # Re-fetch your referral link + stats GET /referral Authorization: Bearer <access_key> # Public interest tally GET /interest -> { total_interest }
Product feed: /agents.json · Site summary: /llms.txt · Referral: 10% both sides, one per customer, 10% max, honored if and when paid ordering opens and an eligible order completes.