NOL / NOLCLI / nolOS
Persistent AI Runtime

NOL

A self-hosted runtime for persistent AI agents with continuity, repair, and re-entry across sessions.

Self-hosted Synthesis-based memory Private runtime Python / Lisp
Current status

Working system in active development

Nevis already runs on a self-hosted stack with an interactive shell, multi-provider adapters, and a mutation pipeline that carries synthesized working context across sessions.

What is NOL

A coordination runtime, not a chat wrapper

NOL is being designed as a language and runtime for durable human-agent collaboration. It focuses on continuity, traceability, bounded action, repair, and return.

What is NOLCLI

Interactive shell and captain bridge

NOLCLI is the operator-facing shell for session continuity, mutation control, runtime inspection, and future command-language work.

What is nolOS

The longer horizon

nolOS is the wider system horizon: control plane, runtime topology, packaging, and environment semantics for persistent agents.

Why it is different

Synthesis instead of shallow retrieval

The core memory line is based on structured re-reading and synthesis of working state, not only similarity search over past snippets.

Stack

Current technical line

  • Python / FastAPI backend
  • Common Lisp / SBCL interactive shell
  • Adapter layer for multiple model providers
  • Self-hosted VPS deployment
About

Alexey Rybnikov

Founder of NOL. Building a self-hosted persistent AI runtime with a strong focus on continuity, architecture, and operator-controlled systems.