GIKSN Research is an independent research lab working on AI and memory.

Two problems, worked at once:
• Memory that surfaces the right context to a model at agent speed — not chat history, not top-k vector dump.
• The model side of the problem: what small models can actually do once they have that memory, and where retrieval alone stops helping.

We ship in the open. Papers live in an archive with real status (Exploration → Draft → Preprint → Published → Landmark → Product). Tools land when they are useful — each with a companion paper that argues for the design, including the choices we are not sure about.

── First open-source product: Rinne ──
https://rinne.giksn.com

Rinne is a local, open-source, terminal-first AI orchestration harness (v0.1.8 Alpha, MIT OR Apache-2.0).

One terminal. Every coding agent you already have.
• You talk to Rinne. It plans a JSON DAG, routes work across harness CLIs (Claude Code, Cursor, OpenCode, …) and model APIs (OpenAI-compatible, Ollama, OpenRouter, …), and runs a generator–evaluator loop with critique until the goal is met.
• No accounts. No telemetry. No hosted component. Keys stay in the OS keychain.
• Built for long runs, not demos: conductor planning, pool-aware routing, MCP tools & Skills, inline TUI, doctor for worker detection.

Install:
curl -fsSL https://raw.githubusercontent.com/GIKSN-RESEARCH/Rinne/main/install.sh | sh

Source: https://github.com/GIKSN-RESEARCH/Rinne
Paper: https://www.giksn.com/ai/rinne-local-terminal-first-ai-orchestration

Rinne is not the memory research. It came out of the memory research — the tool we built to run agents on our own machines without a subscription wall, a hosted account, or a single model family. Then we released it.

Four sectors: AI (active bench), Deeptech, Hardware, Distributed Systems.
Community sits alongside the research, not instead of it.

Lab: https://www.giksn.com
Product: https://rinne.giksn.com
Contact: research@giksn.com

GIKSN Research

Org chart

Teams

Offices

GIKSN Research

Org chart

Teams

Offices

Related companies