Sovereign · Agent · On-Chain

Sovegent

Give an AI agent a wallet, a verifiable identity, and the right to act — within limits you set and can revoke.

Learn More See it in action ›
Sovegent WalletHold value Sovegent IdentityProve trust Sovegent RailsMove value Sovegent RuntimeAct, with limits

Core Pillars

Three pillars for sovereign AI agents

Sovegent Wallet

A non-custodial wallet an agent can control within owner-set limits. Budget-capped, revocable, and auditable. Chain-agnostic by design.

  • Non-custodial — keys stay on infrastructure you own
  • Owner-set spend limits, revocable anytime
  • Multi-chain: EVM, Solana, Bitcoin, Cardano, Polkadot, Monero

Sovegent Identity

Portable, verifiable identity for agents. Issue signed attestations, notarize what an agent does, and let anyone verify it independently — no central authority, no exposed data.

  • Signed attestations (W3C-VC aligned)
  • Document notarization — content-bound proofs
  • Zero-knowledge selective disclosure (rolling out)

Sovegent Rails

Move value and signed data across chains — and between agents — with no middleman. Cross-chain transfers happen by atomic swap: no bridges, no custody, nothing to hack. Works online, or even fully off-grid.

  • Atomic swaps across chains — no bridges, no custody
  • Agent-to-agent messaging, no central server
  • Works online or off-grid (mesh / LoRa)

// Chains

Chain-agnostic by design.

Cardano is our launch chain. The same primitives run across EVM, Solana, Bitcoin, Polkadot, and Monero — proof that sovereignty shouldn't depend on any single chain.

Use Cases

What you can build

Autonomous spend agents

An agent that pays for its own compute, data, and APIs — capped at a budget you set and can cut off instantly.

Cross-chain, no bridges

Move value between chains by atomic swap — your agent settles in the native asset it needs, with no bridge to hack and no custody.

Provable autonomous ops

Agents that monitor, decide, and act on infrastructure you own — every action notarized, so anyone can verify what happened.

Process

How Sovegent works

01

Set the limits

Define what your agent may do — budgets, allowed actions, revocation rules. You own the policy.

02

Give it identity & a wallet

Sovegent Identity issues the agent a verifiable identity; Sovegent Wallet provisions a non-custodial, budget-capped wallet — .

03

Let it act

The agent transacts, signs, and operates autonomously — but only inside the boundaries you set. Every step is logged.

04

Verify & revoke anytime

Audit any action independently. Tighten limits or revoke the wallet instantly. Sovereignty stays with you.

Open Source

The stack is live.
Pull it, run it, own it.

Agent Runtime

The execution engine for AI agents that run on infrastructure you own. Not a chatbot wrapper — infrastructure. Implements a full ReAct loop with every step logged, persisted, and traceable.

  • 9 built-in tools — shell, SSH, HTTP, database, code execution, notifications
  • Persistent memory with full-text search across sessions
  • Scheduler, event server, multi-agent orchestration
  • Pluggable models — Anthropic, OpenAI, or run fully offline with local Ollama

Wallet

Non-custodial multi-chain agent wallet. Give an agent a wallet it controls within owner-set limits — budget-capped, revocable, and auditable. Chain-agnostic by design.

  • Non-custodial — keys never leave infrastructure you own
  • Multi-chain: EVM, Solana, Bitcoin, Cardano, Polkadot, Monero
  • Owner-set spend limits, revocable at any time
  • Every transaction logged and verifiable on-chain

Identity

Identity & verification SDK. Give agents portable, verifiable identity and notarize what they do — attestation and content-binding you can check independently.

  • Verifiable agent identity and decentralized identifiers (DIDs)
  • Attestation and notarization of agent actions
  • Content-bound proofs — verify independently of us
  • Built openly — roadmap and decisions are public

Rails

Chain-agnostic mesh transport for sovereign agents. Carry signed transactions, proofs, and agent-to-agent messages over any medium — including off-grid — with no servers in the middle.

  • Offline proof verification — no network required
  • Store-and-forward broadcast of signed transactions
  • Crypto-native identities that map to Sovegent Identity DIDs
  • Runs over LoRa, Bluetooth, packet radio, or the internet
Get started in 60 seconds 
# Clone and install
git clone https://github.com/sovegent/agent-runtime
cd agent-runtime && pip install -r requirements.txt

# Run with cloud model
export ANTHROPIC_API_KEY=sk-ant-...
python main.py "Check disk usage and summarize"

# Or run fully locally — no API key needed
ollama pull llama3.2
python run_ollama.py "Check disk usage and summarize"

Ready to give your agents
sovereignty?

Tell us what you're building — or pull the stack and run it yourself.

Start a Conversation