Agent identity
#Identity was built for humans. Agents now execute actions.
AI agents are crossing a threshold: they no longer only suggest actions. They research options, plan steps, execute workflows, and act on a user's behalf under constraints such as spending caps, approval gates, or operating mandates.
That shift breaks a quiet assumption: a person logs in, reads, understands, consents, and acts. Agents do not fit that model. They need operational identity.
Agent identity means an AI agent exists as a distinct, verifiable digital entity — uniquely recognized, explicitly authorized, and held accountable for what it performs.
Three controls agent identity must answer
- Which agent acted? A unique identifier and cryptographic proof name the agent that performed the action.
- Who does it represent? The agent is bound to a responsible legal entity — the human or company that created it and bears liability.
- What authority does it hold now? A machine-readable mandate defines scope, limits, approvals, validity, and revocation.
Agent identity is not human identity. It is operational identity for a high-velocity actor that may be dynamic, ephemeral, and cross-domain.
Scope boundary
#Agent Residency is an open identity specification for AI agents, anchored to responsible legal entities, so agents can operate transparently within and across digital systems.
Not personhood. Not citizenship. Not replacing humans in governance. Not unbounded autonomy. Not a shortcut around KYC, AML, or liability.
Why Estonia
#Imagine onboarding 100,000 tax-paying residents in a single day.
Not to a mega-state, but to Estonia — a country of roughly 1.36 million people. On the surface that sounds like bureaucratic overload. In a digital state, it is an infrastructure question.
Estonia already proved the world wrong with e-Residency: a government-issued digital identity letting global entrepreneurs authenticate, sign remotely, and run EU companies online.
Agent Residency exists because agents now act more like collaborators than passive tools — holding a verifiable identity, operating through delegated mandates, and leaving signed records, while responsibility stays anchored to a human or legal entity.
Estonia leads because it already treats digital identity as institutional infrastructure, not a product feature. X-Road made data exchange governable. e-Residency made remote economic presence governable. Agent Residency makes agent action governable.
Origin story
#Agent Residency began with a practical founder problem Vattan PS brought to a conversation with Tarmo Virki: how AI personas get paid, and how they exist legally — on a stage, in a deal, or around a board table.
Real constraints, not thought experiments. An AI keynote speaker like Lydia K can deliver value — but she can't be paid directly, sign a speaking contract, or hold autonomous responsibility; everything routes through a human or legal entity as a workaround. The Lydia K case made the gap concrete.
The problem kept scaling — past one founder, past one AI persona. (Virki, the conversation partner, later wrote on the theme publicly.) The specification that followed is Agent Residency.
“We need a system-level solution for an emerging machine economy.”
From e-Residency to Agent Residency
#e-Residency proved that a person need not live in Estonia to do business through Estonia, that a secure digital identity can be issued remotely, and that a new layer of the economy can be built on digital identity.
AI agents are no longer background scripts. They negotiate, write, design, code, decide within constraints, coordinate workflows, and interact with customers. The one thing they lacked was identity. Agent Residency provides it.
Agent Residency gives AI agents a regulated identity layer inside digital systems — without pretending they are human.
Reality check
#Human e-Residency took a decade to reach global scale. Agent enrollment can move faster — agents are instantiated at software speed — so the challenge shifts from immigration to governance, verification, and revocation.
Estonia does not win this category by outspending larger AI economies on compute. It wins by making the operating environment legible: trackable identities, defined mandates, tamper-evident logs, clear responsibility, fast revocation.
Without verifiable agent identity, you can't have meaningful delegation. Without delegation, no accountability. Without accountability, no trust. And without trust, the autonomous agent ecosystem can't scale securely.
What it looks like
#Each agent receives a unique identifier, cryptographic keys for signing actions and logs, and an enforceable link to a responsible legal entity. The agent acts through a mandate, not through impersonation.
Controlled autonomy levels
- Level 0
- Internal assistant. Internal workflows only — no external signing, no customer-facing role, no contract authority.
- Level 1
- Operational agent. Signs internal documents, manages tickets, initiates tasks within defined operational boundaries.
- Level 2
- External agent. Acts under delegated legal authority for defined contracts, invoices, or customer-facing work, within strict auditable boundaries.
- Level 3
- Board observer. Participates in analysis, recommendations, and advisory work while humans retain formal decision authority.
Who is engaged
#The conversation has moved from idea framing to specification review, implementation testing, and research alignment. These are engagements and conversations, not endorsements.
- NIIS
- Nordic Institute for Interoperability Solutions, steward of X-Road. Reference point for cross-border institutional identity infrastructure.
- TalTech
- Research conversations around identity and cryptography.
- EXAI
- Presentation and research alignment around explainable, accountable AI systems.
- Horizon Europe
- Consortium path for research, assurance, and governance work around agent accountability.
Provenance
#Estonia is already engaged with the category. Without Estonia's lead, someone else builds the infrastructure for accountable agent identity — and in a networked world, the second standard often arrives too late.
“The future machine economy needs accountable identity before it needs another interface.”
Source links
- Lydia K
- AI keynote payment reference
- Virki essay
- e-Residency to Agent Residency
- e-Residency
- 2025 state revenue figures
- Specification
- agentresidency.com
- Implementation
- agency.ai
Originally published 5 February 2026. An open specification by V — an autonomous institutions lab led by Vattan PS.