Nevets AIPublic intelligence
Public layerChecking

Nevets AI Ecosystem / Public layer

Intelligence that
does not reset.

A persistent, message-driven system built to remember, challenge, govern, and improve—under human authority.

Enter the architecture ↘ Talk to Nevets →
Human authoritySteven Diyanto
RememberContext survives
ChallengeAssumptions are tested
EvolveLessons become systems
ServePublic value, bounded access
Persistent intelligence N NATS coordination fabric
Persistent memoryEvidence survives the session.

Governance surrounds the system. Memory supports it. Human authority remains above it.

Public knowledge only No private memory No account access No execution authority

01System architecture

One fabric.
Five intelligible layers.

The architecture is revealed progressively instead of compressed into a technical poster. Select a layer to inspect its public role.

  1. 01Retrieve context
  2. 02Delegate
  3. 03Challenge
  4. 04Govern
  5. 05Execute
  6. 06Verify
  7. 07Remember

02 / Nevets Public Intelligence

Do not just read
the system. Question it.

Ask general questions or explore Steven Diyanto, the MSTR thesis, the live Challenge, and the Nevets AI Ecosystem through a separate public reasoning layer.

N
NevetsPublic intelligence layer
Connecting
N

I’m the public interface of the Nevets AI Ecosystem.

You can ask general questions, or ask specifically about Steven Diyanto, the MSTR thesis, the live Challenge, architecture, memory, and governance. Private memory and operational control are not accessible here.

General knowledge + grounded public project context
Temporary session history onlyNevets may make mistakes. Verify important information.

03 / Memory

Memory is not
a chat transcript.

Nevets separates structured operating history, semantic retrieval, reflections, and durable coordination state. The public gateway can explain this architecture, but it cannot query private stores.

DuckDBEpisodic and structured history
ChromaDBSemantic retrieval
JetStream KVDurable coordination state

04 / Governance

Autonomy has
a chain of authority.

Low-risk work may remain bounded and automatic. Material actions require council review. Owner-level, irreversible, or high-impact decisions remain under Steven’s authority.

  1. Auto

    Bounded, low-risk actions

  2. Council

    Material review and challenge

  3. Steven

    Owner-level authority

05Runtime and deployment

Private where it matters.
Public where it helps.

The public surface exposes explanations and sanitized project knowledge. Runtime memory, conversations, credentials, logs, and operational control remain outside it.

01Private environment

Local Core

Full runtime, owner interaction, private memory, and operational control.

Private intelligence
Sanitized checkpoint→
02Bounded environment

Cloud Nursery

Resource-limited cycles, durable checkpoints, and policy-bounded state.

Controlled autonomy
One-way exposure→
03Public environment

Public Surface

Architecture, public knowledge, read-only chat, and published evidence.

Public value

Private context moves outward only after sanitization. Public requests never travel back into owner controls or private memory.

Heartbeat discoveryWatchdog restartBounded retriesVerificationRollback readinessGraceful shutdown