Neural Commander Neural Commander
← All posts

1 March 2026 · Bradley C Hughes

Why Self-Sovereign AI Infrastructure Matters

Every AI tool forgets everything when the session ends. Neural Commander exists because that's not good enough.

architecture sovereignty vision

Every AI tool you use today has the memory of a goldfish.

Claude doesn’t know what ChatGPT discussed. Your terminal doesn’t know what your IDE planned. Every morning you re-explain the architecture. Every afternoon you re-establish the same decisions.

This isn’t a minor inconvenience — it’s a structural failure in how we interact with AI.

The Dexinator Bug

It started with a trading bot. Dexinator — a decentralized perps DEX terminal — hit a persistent bug. A system crash wiped the Claude Code session, taking with it the requirement context, the architectural decisions, and the debugging progress.

The capital accounting logic was wrong, and the context needed to fix it was gone.

That was the moment Neural Commander stopped being an idea and started being a necessity.

What Self-Sovereign Means

Self-sovereign means three things:

  1. Your hardware. The daemon runs on your machine. Not someone else’s cloud.
  2. Your data. The encrypted vault stays on your disk. Zero telemetry. Zero cloud dependency.
  3. Your rules. You choose which models, which providers, what stays local and what goes to cloud.

It’s not about being anti-cloud. It’s about having the option. The PII encryption gate automatically routes sensitive queries to local Ollama models while letting non-sensitive queries use cloud providers for speed.

Four Memory Layers

The architecture is designed for permanence:

  • Layer 1: Agent Hot Memory — Per-agent isolated context. Each persona has its own soul, tools, and permissions.
  • Layer 2: Cold Memory — All conversations indexed and searchable across agents.
  • Layer 3: Team Hot Memory — Agent swarms with shared context, reaching consensus.
  • Layer 4: Federated Agent Memory — Agents negotiating across Holochain membranes.

The schema designed for Layer 1 determines whether Layer 4 is a clean extension or a painful refactor. We designed for Layer 4 from day one.

What Ships at Launch

Neural Commander v1.0 ships with:

  • Go daemon with persistent memory vault and session intelligence
  • Provider Gateway routing across 9 verified models
  • Telegram bot and CLI for full remote control
  • Email Intelligence with IMAP sync, 4Q classification, PII vault, and sender triage
  • Free forever.

The Community Edition runs entirely on your hardware with zero cost and zero telemetry.

Neural Commander is launching in 2026. Join the waitlist to get early access.

← All posts Join Waitlist