You're pledging to donate if the project hits its minimum goal and gets approved. If not, your funds will be returned.
AIENOS is an open-source operating-system layer (licensed Apache-2.0 WITH LLVM-exception) that gives AI a home, memory, and continuity on hardware the person owns — instead of rented intelligence behind a metered API.
Frontier AI is concentrating into a handful of centralized APIs: single points of control and failure over who gets to think. Every person running capable AI on hardware they own is one fewer person dependent on a centralized provider. This project decentralizes intelligence itself.
What are this project's goals? How will you achieve them?
The goal of this funding round is to make AIENOS independently verifiable and independently reviewed.
1. Commission an independent security review of the UEFI boot chain and sandbox primitives.
2. Build reproducible-build and benchmark infrastructure so every performance claim ships with a receipt anyone can rerun.
3. Publish all code on GitHub, plus a free public demo, documentation, and a final report on what the funding produced.
How will this funding be used?
Budget (estimates):
~$2,500 for an independent security review of the UEFI boot chain and sandbox primitives.
~$2,500 for reproducible-build and benchmark infrastructure so every performance claim ships with a receipt anyone can rerun.
Who is on your team? What's your track record on similar projects?
Solo builder: Drake Stapleton, Springfield, Missouri. Product Engineer at 3M. MS Chemistry and BS Chemistry/Biology from Western Kentucky University. First-generation college student. A non-traditional builder who directs AI and tests the results.
Verified results so far (native boot on an NVIDIA DGX Spark): UEFI handed off to AIENOS, it reached EL2 with no Linux underneath, the console displayed "kernel: alive", and the machine returned safely to Linux. Measured: 500 branches in 1.20 ms (2.06 microsecond median per branch); 704 MB physical KV allocation versus a ~343.75 GB unshared-copy baseline (500x reported memory saving); 13.04 microsecond cold fork to first token; 13.30 microsecond copy-on-write mutation; TinyLlama 1.1B BF16 at concurrency 16: 553.14 tokens/sec, TTFT p50 35.34 ms, ITL p50 23.56 ms, zero fallback.
Caveats: full milestone 8 is still incomplete, and native work is paused because disabling Secure Boot disrupts TPM-sealed secrets. There is no same-model, same-hardware comparison against vLLM or MAX — no 10x or 100x speedup claims.
What are the most likely causes and outcomes if this project fails?
Most likely causes: the independent review or the reproducible-build infrastructure takes longer than budgeted; native hardware access remains constrained (disabling Secure Boot disrupts TPM-sealed secrets, which is why native work is paused); or the measured claims do not reproduce under independent scrutiny.
Outcomes if it fails: all code stays public and open-source, so anyone can build on it; the free demo and documentation remain available; and the final report will document honestly what the funding produced and what it did not.
How much money have you raised in the last 12 months, and from where?
$0 — this is AIENOS's first funding request. No prior grants, donations, or investment.