What does it mean for an AI agent to have an economic identity?

When people talk about "agent identity" today, they usually mean the model or the session — which LLM is running, which API key called it, which thread the conversation lives in. That's an operational identity. It answers "which process is this?" It doesn't answer the question that matters once an agent starts moving money: who is this agent accountable to, and what has it actually done before?

An economic identity is different. It's persistent across models, wallets, and infrastructure changes. It carries a verifiable history — what the agent has transacted, with whom, and under what authorization. And it's portable: swap the underlying model or the wallet provider, and the identity, reputation, and permission history travel with the agent rather than resetting to zero.

That distinction matters because trust between economic actors has never been based on capability alone. A human employee can be capable and still be denied a corporate card, or given one with a $500 limit instead of $50,000. The limit isn't a judgment about their intelligence — it's a judgment based on identity, track record, and role. AI agents that transact need the same layered trust, and today almost none of them have it.

Concretely, we treat economic identity as four connected pieces: who the agent is (a persistent identifier independent of the model running it), who it represents (a person, a company, or another agent), what it's authorized to do (spending limits, approved counterparties, approved contracts), and what it has done before (a verifiable transaction history other systems can check before deciding to trust it).

Get that right, and a lot of downstream problems get simpler. Counterparties can decide whether to transact with an agent based on its history instead of blind trust. Spending limits can be enforced at the identity layer instead of bolted onto each integration separately. And when something goes wrong, there's an audit trail tied to a specific, accountable identity — not just a wallet address or an API log line.

This is the layer AdaSouls is building: persistent identity, verifiable reputation, and programmable economic authority for AI agents, independent of the model, wallet, or framework running them.