2 Identity & Trust

Who is the agent?

The most unsolved problem in agentic payments. Before a payment system allows a transaction, it needs to know: is this a human, a bot, or a legitimate autonomous agent? That answer doesn't exist yet.

Layer 2 establishes the identity of the entity making a payment — authentication, KYC, bot detection, and increasingly "Know Your Agent" (KYA). Without this layer, every other payment control is bypassed.

The Unsolved Problem

Every payment system built in the last 50 years assumes a human is at the keyboard. KYC collects a passport scan. 3DS sends a one-time password to your phone. Fraud models look for "human-like" behavior. Agents fail all of these tests by default — they move too fast, from too many IPs, with too-perfect timing. The industry hasn't agreed on what "prove you're a legitimate agent" even means.

Companies building this layer

Skyfire
Agent Payments

Skyfire gives AI agents a payment identity — a spending account with programmable controls. Agents get a Skyfire-issued credential that payment systems recognize as a legitimate, policy-bound entity. Backed by major VCs; one of the first companies purpose-built for the agent payment layer.

ModelAgent spending accounts + identity credentials ControlsSpend limits, merchant allowlists, per-agent policies StageLive, raising / scaling
skyfire.xyz →
KYAPay
Know Your Agent

KYAPay (kyapay.org) addresses the "Know Your Agent" problem — verifying that an AI agent is what it claims to be, operating within sanctioned limits, and authorized by the human or organization behind it. Described by Apify as a must-have element for any agentic toolkit.

FocusAgent identity verification + authorization proof EcosystemConnected to Skyfire's agent payment network SignalHighlighted by Apify, backed by Skyfire team
kyapay.org →

Supporting infrastructure (adapted from human identity)

Forter

Fraud Prevention

Real-time fraud decisioning using behavior + device signals. Built for human transactions but increasingly used to flag anomalous agent behavior.

DataDome

Bot Detection

Detects and blocks malicious bots at the edge. The irony: legitimate AI agents often look identical to the bots DataDome blocks. No "allowlist for good bots" yet.

Socure

KYC / Identity

Automated KYC and identity verification for onboarding. Designed for humans (face match, document scan) — agents have no face or document to scan.

Sardine

Compliance + Risk

Fraud and AML platform with strong crypto coverage. One of the better-positioned companies to extend into agent identity as the space matures.

Persona

Identity Verification

Flexible KYC/KYB platform that could be extended to "KYA" — Know Your Agent — with the right policy primitives.

Privy

Embedded Wallets

Gives agents a cryptographic identity via embedded wallets. Not identity verification per se, but a stepping stone: an agent with a consistent on-chain address is at least accountable.

The Gap: No Standard for "Proof of Agent"

The industry needs a "proof of agent" credential — something that says: this is an AI agent, authorized by this human/org, with these spending limits, operating in this context. It needs to be machine-readable, cryptographically verifiable, and accepted by payment processors without a human loop. Skyfire and KYAPay are early attempts. The winner of this layer may be the next Stripe — the identity infrastructure everyone builds on top of.