Erebor Research System

Private intelligence for the AI infrastructure decade.

Erebor is Joshua's private research and portfolio-intelligence workspace: built to study compute, energy, semiconductors, industrial capacity, and the companies that may matter as AI becomes an infrastructure race.

View product direction Security posture

Core thesis

AI progress becomes a bottleneck map.

The important questions move beyond model demos: who can secure power, chips, memory, cooling, interconnects, factories, grid access, and durable operating discipline?

Compute supply chainsPower and grid bottlenecksSemiconductors and packagingIndustrial AI infrastructure

0private records exposed

Current web shell is intentionally public-safe and contains no raw X data, credentials, ledger data, or local paths.

Product roadmap

Two products, cleanly separated.

Erebor becomes the research cockpit. Demerzel becomes the operating-system surface. They can share design language without mixing repositories, credentials, or private datasets.

Now

Public shell

A safe, visible Vercel entrypoint with no private data and no live trading surfaces.

Authenticated cockpit

Clerk-protected research workspace, Joshua-only allowlist, sanitized read model, and audit log.

Later

Action queue

Request research, promote ideas, and approve bounded Demerzel actions without exposing local systems.

Initial lanes

AI power stack

Data-center equipment

Memory and advanced packaging

Nuclear, fusion, and grid hardware

Safety boundary

Public website first. Private cockpit only after auth.

This deployment is a visible foundation, not the private research engine. The next layer will require Clerk authentication, a Joshua-only allowlist, a sanitized cloud read model, and an audited action queue before any operational capability is exposed.