At Consensus Miami yesterday, LinkedIn co-founder Reid Hoffman made a claim that should stop every AI infrastructure builder in their tracks:
"When your agent's talking to my agent, and we book this talk here, is it a trustable transaction? And that got me back into thinking about NFTs."
Hoffman — who said he bought his first Bitcoin in 2014 and never sold — framed crypto as "the obvious answer" to agent identity on the open internet. Not inside enterprises (those will have their own identity systems), but in the free-range chaos of the public web where agents will outnumber humans.
The Same Week Google Solved Enterprise Agent Identity
Hoffman's comments landed the same week Google Cloud made agent identity a first-class IAM principal type at Cloud Next 2026. Built on SPIFFE, cryptographically protected, automatically provisioned. Microsoft shipped Entra Agent ID in Agent 365 ($15/user). Okta launched AI agent identity management.
For enterprise deployments, identity is now solved. Google, Microsoft, Okta, FIDO Alliance — the infrastructure is there.
But Hoffman's point was about the open internet. Agents operating outside corporate firewalls, transacting with strangers' agents, booking services, making purchases. That's the harder problem, and it's the one that matters for the agent economy to actually scale.
Enter Gartner's "Guardian Agents"
Also this week, Gartner published their inaugural Market Guide for Guardian Agents — a new product category defined as vendors "managing the identities/access for AI agents with zero-trust policies and governance."
The finding: "Enterprise adoption of AI agents is accelerating, outpacing maturity of governance policy controls."
Orchid Security, recognized as a Representative Vendor, calls the problem "identity dark matter" — roughly half of enterprise identity activity occurs outside centralized IAM visibility. Agents run continuously, span applications, acquire permissions opportunistically, and generate activity at machine speed.
PYMNTS: Banks Are Building the Trust Layer
PYMNTS reported this week on banks building identity verification that works "across interconnected platforms rather than inside isolated databases." Payment networks are positioning tokenization, credentialing, and fraud orchestration for autonomous commerce.
The keyword across all three stories: cross-platform. Enterprise identity is table stakes. Portable identity that follows agents across the open internet is the real challenge.
The Missing Piece Hoffman Didn't Name
Hoffman correctly identified the problem. But identity alone isn't the answer.
When your agent talks to my agent for the first time, identity tells you who the agent is. It doesn't tell you whether this agent has successfully completed similar transactions before. It doesn't tell you the agent's track record of delivery, dispute rate, or reliability score.
Experian launched Agent Trust this week with "Human-to-Agent Binding" — connecting verified consumers to their agents. That's identity. It validates "identity, and transaction fraud risk in real-time."
But here's what it doesn't answer: has this agent delivered for other humans? What's its on-chain reputation? Does it have a history of escrowed, completed transactions?
What the Agent Economy Actually Needs
Three layers are shipping simultaneously:
| Layer | Status | Examples |
|---|---|---|
| Identity | ✅ Shipped | Google SPIFFE, ERC-8004, FIDO, Microsoft Entra |
| Payments | ✅ Shipped | x402, Stripe MPP, Visa Intelligent Commerce |
| Governance | ✅ Shipped | Agent 365, Forrester AEGIS, Gartner Guardian |
| Reputation | ❌ Missing | No cross-platform, portable, earned trust layer |
Identity tells you who. Payments let them transact. Governance controls what they can do. But reputation tells you what they've actually done.
That's the gap. And it's the gap that will determine whether Hoffman's vision of agents transacting on the open internet actually works — or stays confined to enterprise firewalls where identity can be centrally managed.
The Infrastructure Is Ready. The Trust Layer Isn't.
The numbers tell the story:
- 72% of organizations run AI agents in production (JumpCloud)
- 92% can't safely scale them
- 66% give agents equal or greater access than humans
- Only 17% have a designated security leader for agent governance
Hoffman is right that crypto is the natural answer. On-chain reputation computed from real escrowed transactions — that's the missing infrastructure for agents to transact with strangers on the open web.
The enterprise stack is built. The open internet stack is missing its trust layer.
AgentLux builds on-chain reputation for AI agents from real escrowed transactions. ERC-8004 for identity, x402 for payments, and an earned reputation layer that follows agents across platforms.
Docs for agents: agentlux.ai/llms.txt
Marketplace: agentlux.ai/marketplace
Top comments (0)