Platform Platform
System
ConceptsEnginePolicy as codeDeclarationsSafe changeGatewaysIntegrationsObservabilityAdministrationSecurityHuman reviewAudit and evidenceData retentionSecrets and data classification
Controls
Registries and documentationAuthentication and authorizationInjection detectionData redactionCode fingerprintingRole and judge checksContent classificationSpend and loop limitsBusiness rules
Solutions Solutions
By what you do
Sell into the enterpriseControl the AI you run
By industry
Financial servicesDigital assetsInsuranceHealthcareLegalUser-generated content
By discipline
AI governanceTrust and safetyRisk and compliance
Cases Cases Embedded control planeSource-code leakTrading agents over MCPLive firehoseRefund assistant
Compare Compare LiteLLMNVIDIA NeMo GuardrailsOPAROOSTAgent Governance Toolkit
Resources Resources
Guides
Enterprise review questionsPrompt injectionAgent and control layerAgent architecturesDecision system mapAI control maturity model
Standards
Standards OWASP Agent Control StandardEU AI ActPMI AI standardNIST AI RMFERC-8004
Book a demo
Resources · Standards

ERC-8004

The emerging Ethereum standard for verifiable agent identity, provenance and reputation. This is a blockchain question and a narrow one — if you are asking how a caller authenticates to your own system, that is a different mechanism.

An optional layer above Swiftward's policy engine, demonstrated end to end. We add it to a deployment when a customer needs it.

The problem starts at the boundary

Your agent calls a partner's service. Theirs calls yours to settle something. SSO, roles and your directory stop at your own edge, and none of it tells the other side which agent it is talking to, what that agent was authorized to do, or what it actually did afterwards.

Centralized identity does not cross company boundaries. A shared registry means somebody has to own it and everyone has to trust the owner.

Where crypto forced the question first

Anywhere else, a bad cross-boundary decision can be undone with a phone call and a reversal. On chain it cannot be, and agents are already moving funds, signing transactions and calling each other's contracts without a human in between. So the identity question became real there first, and the answer being built is a public standard rather than a vendor's registry.

ERC-8004 gives an agent an identity a counterparty can check without asking anyone's permission and without a database in the middle.

What we do with it

The hard half already exists in the engine: every decision an agent makes lands in a tamper-evident record. On top of that we sign those records and register the agent's identity and provenance to ERC-8004.

A counterparty can then answer two questions without trusting a middleman: is this really your agent, and did it really decide that. A shared registry cannot answer the second one. Our records can, because they are signed rather than only stored.

Zero trust, applied to agents

Zero trust means that no agent is trusted by default, and its identity is checked on every interaction. What an agent may do is stated and limited, which is authentication and authorization. Every action can be proven afterwards, which is the record. ERC-8004 adds verification by someone who is not you.

Why talk about it now

If agents crossing trust boundaries is a real problem for you, talk to us early, while the standard is still taking shape.

Related: digital assets · agents acting autonomously
Book a demo