Get started

Pick what you want to do. We'll show you how.

I want to run models on my hardware

You have a GPU and you want to run AI models locally. No cloud, no accounts, no fees.

nollama — 5 min setup

I want my AI to remember things

You're already using an AI tool and want it to have persistent memory that you control.

seidr — works with Open WebUI, Open Code, and more

I want to understand the protocol

You're building an agent framework or memory backend and want to implement A2M.

A2M Protocol — interactive explorer

nollama

Run models on your hardware

If you've used Ollama before, this will feel familiar — but without the Docker container, the account requirement, or the opaque blob store. If you haven't, this is the easiest way to run AI models on your own machine.

What you need

A computer with a GPU. NVIDIA (any modern card with 6GB+ VRAM), AMD, or Apple Silicon. That's it. No Docker, no Python, no special accounts.

Install nollama

# Download and install$ curl -fsSL https://get.nollama.dev | sh# Or with your package manager$ nollama runtime install

Pick a model

Models come from HuggingFace. No registry accounts needed — just a direct download. Here are some good starting points:

Model VRAM needed Good for

Qwen3-8B-Q4_K_M 6 GB General chat, coding assistance

Qwen3-14B-Q4_K_M 10 GB Better reasoning, longer context

Qwen3-32B-Q4_K_M 20 GB Complex tasks, code review

Start chatting

$ nollama pull unsloth/Qwen3-8B-GGUF:Q4_K_M$ nollama serve && nollama load Qwen3-8B-Q4_K_M.gguf✓ Model loaded on GPU 0 — port 11435✓ OpenAI-compatible API ready at http://localhost:11434

That's an OpenAI-compatible API. Any tool that works with OpenAI's API works with nollama. Point Open WebUI, Continue, Cursor, or any other client at http://localhost:11434.

seidr

Give your AI memory

Most AI tools forget everything the moment you close the conversation. Seidr gives them persistent memory — stored in your database, on your hardware, under your control. It works with any tool that supports MCP (Model Context Protocol) or OpenAI-compatible APIs.

What you need

Docker (or Podman) and PostgreSQL. Seidr handles the rest — embeddings, vector search, the memory protocol. If you're already running nollama, Seidr can use it for embeddings too.

Start Seidr

# Pull and run$ docker run -d --name seidr \  -p 8200:8200 \  -e DATABASE_URL=postgresql://user:pass@host:5432/seidr \  -e EMBED_URL=http://localhost:11434 \  ghcr.io/sovereignty-labs/seidr:latest# Verify it's running$ curl http://localhost:8200/a2m/health✓ {"status": "healthy"}

Connect your tools

Seidr exposes an MCP endpoint. Any MCP-compatible client can connect directly.

Open WebUI

Settings → Tools → Add MCP Server → http://localhost:8200/mcp

Open Code / Claude Code

Add to your MCP config: "seidr": { "url": "http://localhost:8200/mcp" }

Any OpenAI-compatible client

Seidr's A2M endpoint works with any HTTP client. POST /a2m with JSON-RPC payloads.

Your AI now remembers

Once connected, your AI assistant can use remember and recall tools automatically. Memories persist across sessions, grow in confidence with validation, and stay on your hardware.