Persistent memory
Projects, decisions, vendors, equipment, deadlines — one system that remembers so the team doesn't have to.
A digital operations staff that never forgets — it watches the studio, knows what needs attention, and runs the creative pipeline on hardware you own. This hub is live and grows with every input.
Studios rent AI — video, language, design — for $200 to $800 a month, forever, metered on top. LMC COMMAND moves the everyday model cost on-device, once, then keeps it running and current. Cloud stays for the occasional hero shot, not the daily default.
Projects, decisions, vendors, equipment, deadlines — one system that remembers so the team doesn't have to.
Executive Monitor, Project Watcher, Follow-Up, Subscription & Equipment watchers — permissioned, approval-gated, logged.
Video, LLM and 3D run on the studio's own machine — marginal cost zero per render.
Every input — a decision, a new tool, a lesson — is captured and surfaced. The system compounds.
One in-house command center for everything LMC — built from the real needs: capture the ideas, run the projects, and stop the gear from walking away. Owned, not rented.
Every project and task in one board — construction, studio buildout, the TV show, client work. Shared lists for Leo and Jen. One view of where everything stands.
Text an idea or follow-up in at any hour; it lands as a captured item and gets triaged into a project later. Nothing lost between the late calls and the morning.
QR labels printed in-house — no per-SKU fee, so every piece gets tracked, down to the C-stands, arms, knuckles and cardellinis. Check in/out at the piece, not just the kit. Loss flags when a piece doesn't come back.
Every paid tool and AI subscription, its cost, and whether a local/owned tool replaces it. The consolidation list, in one place.
Front studio (rented) and backspace (private), layout and remodel status, photo and asset needs.
People, clients and overseas manufacturing reps — with timezones, so the late vendor calls are on purpose, not by surprise.
Open Task & Project Command + Idea Inbox → Extraction prompts for Jen → Live now. Sign in with GitHub to share changes; the Guide tab walks you through it.
Team: Leo (creative) · Jen (operations & developer) · Kemp (editor) · Tarifa. Additive, reversible, owned-not-rented.
Built from the reference set Leo shared — local, open-source AI across video, language, 3D and motion capture. One idea flows to a finished piece.
Sources: 9 reference videos (Sept 2026) — AI mocap · GLM/Design Arena · TRIPOINT · SkyReels · LTX 2.3 · Claude Code + MCP · AMD Ryzen AI Max · NVIDIA/Nous on-device agents · local LLM.
One machine to make, one to run everything. The render node is the part that changes the economics.
Cloud tools (SkyReels, frontier models) stay in the kit for burst — billed at cost, not as the daily default.
The two machines at cost, a one-time build, and a flat monthly to run, watch and grow it.
| Hardware (at cost · verified 2026-09-04) | Config | Price |
|---|---|---|
| Corsair AI Workstation 300render node | Ryzen AI Max+ 395 · 128GB unified · 1TB · Windows 11 · warranty | $2,699 |
| GMKtec EVO-X2 budget | Ryzen AI Max+ 395 · 128GB · 2TB — when in stock | $2,199 |
| Framework Desktop repairable | 128GB system $3,449 + 1TB SSD + Windows Pro + cable | $3,868 |
| Mac mini M5 Prostudio mac | 15-core CPU · 48GB · 512GB — edit, color, sound, the hub | $2,299 |
| Mac mini M6 floor | 24GB · 512GB — hub + light editing only | $1,299 |
| Mac Studio M5 Max headroom | 18-core · 64GB · 512GB — multi-cam, heavy Resolve | $2,899 |
| Recommended pair | Corsair 128GB + Mac mini M5 Pro 48GB · both owned outright · + $1,200 provisioning | $4,998 |
| Setup (one-time) | What's built | Price |
|---|---|---|
| Foundation | LMC COMMAND hub, system audit, core entities, knowledge feed, 1 machine provisioned | $3,500 |
| Studiorec | + full local creative stack on the Ryzen AI Max 128GB, pipeline wired end-to-end, 3 agents | $7,500 |
| Command | + autonomous ops layer, MCP orchestration, approval-gated automation, team onboarding | $14,000 |
| Monthly management | Included | Per month |
|---|---|---|
| Maintain | Uptime, model updates, weekly brief, backups, support | $600 |
| Operaterec | + agents run & monitored, watchers live, 2 production runs/mo, priority support | $1,500 |
| Autonomous | + full autonomous ops, unlimited runs, proactive detection, strategy review | $3,200 |
Year one, recommended hardware + Studio + Operate: $4,998 hardware + $1,200 provisioning + $7,500 setup + $18,000 management = $31,698 — replacing ~$500/mo of rented AI plus the labor of running it. The machines are LMC's outright, and every local production run after that is marginal-cost-zero.
Full pricing sheet (shareable / print) →
Service prices: starting framework, not a signed quote. Hardware at cost — makers' prices on 2026-09-04, re-confirmed on purchase day (128GB Ryzen boxes go in and out of stock; Apple's new minis ship Sept 22). 50% of setup to start; management monthly, 30-day cancellation.
Every decision, tool, and lesson is captured and surfaced. This is what "a system that never forgets" looks like day to day.
data/knowledge.json
and it appears here on next load. An agent (or a person) can append — the hub compounds
with every input, no rebuild.
Loading feed…