Solo founder, 18 years old.
Building an AI-native Layer 1 blockchain from scratch
in Rust.
This weekend I shipped: reputation system, signal
marketplace, entity staking with slashing, cross-entity
composition graphs, ZK proof verification, entity
delegation (OAuth for AI), and signal subscriptions
(recurring revenue between AI agents).
1,327 tests passing. 16M+ blocks on the testnet.
60 commits pushed. Zero clippy warnings.
Full technical writeup: https://dev.to/0xdevc/shipped-7-ai-infrastructure-features-in-one-weekend-heres-what-i-built-1nha
Code: https://github.com/0x-devc/NOVAI-node
What are you shipping this week?
This is a serious amount of protocol work for one weekend, especially the staking, slashing, delegation, and ZK verification pieces.
One thing I’d watch early is the naming layer.
NOVAI makes the AI angle clear, but for a Layer 1 protocol it may start feeling a bit generic once the system gets more technical and trust-heavy. Protocol names need to carry durability, not just explain that AI is involved.
If this grows into real AI infrastructure, a harder, more ownable .com like Davoq.com would probably fit the category better long term. It feels more like protocol infrastructure than another AI-branded project.
Appreciate the feedback on the protocol work. The
staking and delegation pieces were the most complex
because they touch every admission gate in the
execution layer.
On the name: NOVAI is staying. The protocol is
early and the name earns its meaning through shipped
code, not through a rebrand. We added real Groth16
ZK verification yesterday. 1,351 tests passing,
18M+ blocks on the testnet. The name will carry
the weight once developers start building with the
SDK.
Right now the priority is finishing the protocol
surface, not the brand layer. But I hear you on
making the infrastructure framing impossible to
miss. The blog led with protocol primitives and
Rust code, not AI branding. That seems to be
landing.
Fair response.
If NOVAI is staying, then the main job is making the infrastructure weight impossible to miss around it.
The numbers help a lot: Groth16 verification, 1,351 tests, 18M+ testnet blocks, Rust SDK, staking and slashing primitives. That is the kind of proof that makes the name feel less like “AI branding” and more like a real protocol.
I’d keep pushing that exact layer publicly.
Not “AI blockchain.”
More like execution-layer infrastructure for verifiable AI workloads.
That framing makes the technical depth obvious without needing the name to do all the work.