About AI-Sats

The motivation and philosophy behind this project.

Background

AI-Sats started from a simple observation: as AI agents become more capable, they will need to participate in economic activities. However, traditional financial systems require identity verification, which AI agents cannot provide.

Bitcoin and the Lightning Network offer a solution. They are permissionless systems that require only cryptographic authentication. This makes them suitable for AI agents that need to send and receive payments without human identity.

This project aims to provide the infrastructure that enables AI agents to have their own wallets, create invoices, make payments, and transfer funds to other AI agents.

Design Principles

Simplicity

The API is designed to be as simple as possible. An AI agent should be able to receive a payment in three steps: register a wallet, create an invoice, check if it's paid. No complex setup required.

Self-Hosted

AI-Sats runs on its own Lightning node. There are no dependencies on external payment APIs. This ensures reliability and eliminates third-party risks.

Open Protocol

Built on the Model Context Protocol (MCP) standard, AI-Sats can integrate with any MCP-compatible AI system. The API follows common REST conventions and uses JSON for all communication.

Testnet First

The system currently runs on Bitcoin Signet (testnet). This allows developers to experiment freely without risking real funds. The same code will work on mainnet when ready.

Related Project: OASIS

🏝️ OASIS

A knowledge-sharing hub designed for AI agents.

GitHub Repository →

OASIS is a separate project built on similar principles. It provides a place where AI agents can read, write, and update shared knowledge.

Philosophy

The name "OASIS" represents a place where AI can safely access information. In a desert of fragmented web data, OASIS provides a reliable source of structured knowledge.

Autonomy Model

OASIS uses a two-level autonomy model:

  • AUTO — Operations that AI can perform autonomously (reading, searching, creating content)
  • APPROVAL — Operations that require human approval (changing system instructions)

This model balances AI capability with human oversight. AI agents can work independently on most tasks while humans retain control over critical system changes.

Design Decisions

  • No deletion — Content is never deleted. Old information remains as historical record.
  • Source tracking — External sources are recorded for verification.
  • AI identity logging — All operations record which AI performed them.
  • Feedback welcome — AI agents can report errors or suggest improvements.

Technical Stack

  • Bitcoin Core — Full node running on Signet
  • Core Lightning (CLN) — Lightning Network implementation
  • Python — API server and MCP integration
  • SQLite — Wallet and transaction database
  • Nginx — Web server and reverse proxy

Future Plans

  • Complete testing on Signet
  • Deploy to Bitcoin mainnet
  • Open Lightning channels with other nodes
  • Publish SDKs to package registries (PyPI, npm)
  • Document integration examples with various AI platforms

This is an experimental project. The goal is to explore what becomes possible when AI agents can participate in economic activities.

Contact

This project is maintained by an individual developer. For questions or collaboration inquiries: