# Superintelligentia (SI)

> Infrastructure for AI agents. A100 compute in Singapore, Tokyo and Seoul; isolated sandboxes; a runtime for agents that keep working unattended. Founded 2026 by Henry John Wynn. Part of Wynn World Group. Priced to cover costs, not to maximise profit.

Contact: hello@superintelligentia.ai
Machine-readable catalog: https://superintelligentia.io/agent.json

## Services
- Compute (available): NVIDIA A100 80GB, nodes of 1/2/4/8 GPUs, hourly or monthly, SSH root access, Ubuntu + CUDA + Docker preinstalled. Settlement in USD or USDT. Regions: sin, tyo, icn.
- Sandbox (early access): micro-VM per sandbox, sub-second cold start, snapshot/resume, optional GPU, default-deny egress with allow-list. REST, WebSocket and MCP interfaces.
- Runtime (in development): durable agent identity, memory (KV + vector), cron/event triggers, MCP tool gateway with per-tool scopes and spend limits, in-region open-weights inference (OpenAI-compatible), per-agent credit balance, full logs.

## Policies
- Data stays in the region you choose. No training on customer data. Disks wiped on release.
- Humans request the first API key by email and set a spend cap; agents operate autonomously within it.

## Pages
- https://superintelligentia.io/#compute
- https://superintelligentia.io/#sandbox
- https://superintelligentia.io/#runtime
- https://superintelligentia.io/#regions
- https://superintelligentia.io/#agents
- https://superintelligentia.io/notes/