About

Why this exists, what I build, and how the Local Vault path changes the risk equation.

The Problem

Most AI infrastructure assumes always-on connectivity. When that assumption breaks — a dropped link, a maintenance window, a security incident — the systems stop. The data they leaked in the meantime isn't a configuration mistake. It's an architecture decision. Audit comes, and nobody can prove what their systems did — or didn't — send.

The industry built for always-on connectivity. The real world isn't always on. Systems that depend on the cloud to think, to verify, to function — these fail at the worst possible moment. And the data they leak in the meantime isn't a configuration issue. It's an architecture problem.

The Position

Integrity — the system does exactly what it claims. No hidden behavior, no secret telemetry, no "trust us." Provable. Auditable. The logs show what happened, and nothing happened that isn't in the logs.

Loyalty — the system stays working for you. Not for its vendor. Not for the cloud. When the connection drops, it keeps running. When the API changes, it keeps running. When the vendor pivots, it keeps running. Offline by design, cloud-optional by choice.

Honesty — you can verify everything. Telemetry transparency: see the traffic, verify the boundary, and prove what did or did not leave. Behavior transparency: inspect what the agent does, not just what it says. Trust measured in evidence, not faith.

These aren't values on a poster. They're testable properties of a system. A system either sends data home or it doesn't. A system either runs offline or it doesn't. A log either shows what happened or it doesn't. You can't fake these with a configuration toggle. They have to be built in.

The Path

Research — Publish the evidence first. The hyperscaler telemetry report is the advance scout: a public map of cloud dependency, logging, jurisdiction, and verification gaps.

Consultancy — Turn concern into evidence. Audit an AI deployment, identify telemetry and resilience gaps, and define what must stay local, what can remain cloud-connected, and what needs stronger controls.

Product — Local Vault is the core package: owned hardware plus local-first AI services for organizations that need continuity, privacy, and auditability. RixBot is the included interface inside that vault, not a separate product line.

Platform — The longer path is a repeatable standard for resilient, private, verifiable AI deployments: auditable where it matters, interoperable where it helps, and honest about every boundary.

Each lane feeds the next. Research earns attention. Audit finds the real requirements. Local Vault turns those requirements into working infrastructure. Build. Improve. Repeat.

The Builder

elect-rix is a solo operation running a two-workstation AI lab. Primary: Pop!_OS with an RTX 5070 Ti for local LLM inference. Secondary: Linux Mint with dual GPUs for distributed builds and compute. The lab exists to test what can be owned, self-hosted, and verified before asking a client to rely on it.

Two connected workstation nodes — primary and secondary AI lab setup

I use AI as a collaborator, not a replacement. The models handle the boilerplate, the research, the repetitive tasks. I handle the architecture, the edge cases, and the judgment calls. That's the only way to build systems that actually work — and keep working when conditions change.

Get in Touch

Building something that needs to work when the cloud doesn't? Want to know whether the hyperscaler report maps onto your own AI deployment? Let's talk.

Start a Conversation