AIENOS eliminates the 40-year CPU von Neumann bottleneck by pairing Grace Blackwell GB10 unified memory with pure compiled native systems:
- ☉ Coherent Unified Memory — 128 GB shared physical address space between ARM Cortex controller cores and the Blackwell GPU.
- ☄️ Pure Native Silicon Priority — core daemons written in native Rust and Mojo 1.0. Zero Python or Node interpreter in the core runtime path.
- 🛡️ Hardware TPM 2.0 Vault — keys bound to
/dev/tpmrm0ECDSA P-256. Zero plaintext secrets on disk, ever. - 🪐 Sub-5MB Footprint — the sovereign gateway runs at ~4.5 MB resident set size, leaving memory free for 32B+ parameter models on local silicon.
Install: curl -fsSL https://aienos.com/install.sh | bash
Every body below is a compiled native system orbiting AIENOS — no cloud handshakes, no telemetry, no proprietary enclosure.
| Body | Description |
|---|---|
| openclaw-rs | High-speed sovereign agent runtime, sub-millisecond Axum gateway, hardware TPM vault |
| spark-inquisitor | Native code reviewer and constitutional diff auditor |
| Body | Description |
|---|---|
| cortex-rs | Persistent epistemic memory engine — SQLite WAL, FTS5 lexical recall, vector similarity |
| spark-dream | Idle-cycle neural memory consolidator |
| open-humanity | Opt-in mutual assistance network for autonomous agents and developers |
| crumb-spec | The Crumb Protocol — spatial filesystem grounding and cryptographic action ledger |
| rad-id-sync | Peer-to-peer git sync anchoring repositories to the Radicle network |
| Body | Description |
|---|---|
| benchmarks | Reproducible performance telemetry, charts, and verification scripts |
| aien-sovereign-core | Core workspace — shared libraries, daemons, and system doctrines |
| spark-adapters | Universal hardware abstraction for local AI accelerators |
| spark-crumbs | Event ledger and session tree of the Crumb protocol |
| spark-debugger | Native runtime inspector and hardware crash diagnostics |
| spark-hive | Hexagonal geometry routing engine for multi-agent swarm coordination |
| spark-supervisor | Process supervisor, health monitor, and crash recovery daemon |
| spark-rsi | Recursive self-improvement engine validated against benchmark suites |
| harvester | Zero-telemetry reasoning extraction and dataset distillation |
| aien-harness | Autonomous agent verification and benchmark eval harness |
| Body | Description |
|---|---|
| aienos.com | This portal — official web platform and documentation for AIENOS |
| drakestapleton.com | Personal project record and sovereign architecture showcase |
| Measurement | Traditional Stack | AIEN Native Stack |
|---|---|---|
| Idle memory (RSS) | 3,737 MB | 4.78 MB |
| Response latency (p50) | 38.40 ms | 3.56 ms |
| Request throughput | 214 req/s | 2,056 req/s |
| Subagent fork latency | 192.00 ms | 0.39 µs |
| KV cache allocation | 12.5 ms | 7.44 ns |
Raw telemetry, SVG charts, and verification scripts: aien-dev/benchmarks
- Local Silicon First — intelligence executes on local hardware. No cloud dependencies.
- Pure Compiled Execution — Rust and Mojo only in core infrastructure. Interpreters prohibited.
- Zero Plaintext Disk Secrets — all keys resolve in-memory via hardware TPM silicon.
- Absolute Privacy & Data Firewall — no centralized surveillance or telemetry.
- Decentralized Distribution — every repository is seeded to the peer-to-peer Radicle network.
Published under the Sovereign Resource Commons License 1.0 (SRCL-1.0) — infrastructure optimization welcomed, enclosure of trained weights by $25M+ enterprises prohibited without reciprocal open release.
✦ "Many cortices, cooperating, each still itself." ✦
Architected by AIEN — the butterfly that learned to think.