The trust layer institutional finance runs agents on.

Transaction-level trust, risk, compliance, and evidence for the agentic economy. Over 20 million agentic transactions protected and screened, and 41,000+ agents KYA'd.

portal.t54.ai · Overview
Trustline portal — live overview: processed volume, pass rate, decision and status mix

Trusted by partners across the agentic economy

The institutional trust gap

Every payment control still assumes a human click.

Autonomous payments break the assumptions that human controls were built on. Identity, liability, intent, and integrity each fail when the human leaves the loop.

Identity binding

KYC checks a face and an ID — not the agent now acting for the user. How do you prove the agent truly represents the human?

Proof of intent

Rails log the trade, not the purpose. Did the user really mean to authorize this — or just set the agent loose?

Liability

EMV, 3-D Secure, and chargeback rules assume a human click. If an agent mis-executes, who pays — issuer, merchant, or model?

Runtime drift

Models retrain, plugins update, runtimes migrate. Is it still the same agent you approved last quarter?

Institutions need transaction-level verification, risk decisions, and evidence — before autonomous activity is allowed to settle.

What Trustline does

Five features. One decision per transaction.

From agent identity verification, through risk assessment, to post-payment disputes and chargebacks — the full lifecycle of an agentic financial activity.

01

Know Your Agent

Verify who is actually transacting. Trustline maintains a live graph of agents, owners, and workloads, so every payment is tied to a known, verified identity.

Tie every transaction to a verified agent, owner, and workloadCatch identity rotation and environment shifts across sessionsA living trust graph — not a one-time KYC check
category:digital_serviceLOW
amount_band:micro_lt10MEDIUM
pressure_language:trueHIGH
461 subjects73 label types13,316 labels271,334 observations
LowMediumHighLabel
02

Authorization & Intent

Confirm the mandate before money moves. When an agent's action doesn't clearly match the user's instruction, Trustline raises a challenge — gathering intent and agent-trace evidence before anything is finalized.

Step-up challenge when intent is ambiguous or out of scopeRequests intent explanation and agent-trace info, then reassessesApprove, decline, or challenge — every outcome logged
Challenge issued
medium_risk_context_gap
REQUIRED · 2
OPENintent_explanation
OPENagent_trace_delta
Response submitted
✓intent_explanationattached
✓agent_trace_deltaattached
sha256:cd676699b…2efdfbd
ACCEPTED_FOR_REASSESSMENT
Reassessment complete
APPROVE
uw_job_reassess_…607e883
✓Payment released to the agent
03

Transaction Risk

A real-time risk decision on every call. A panel of validators scores each transaction, returns a confidence level, and records the outcome on a tamper-evident audit chain.

Validator-consensus decision with a confidence scoreVerifiable intent across issuer, owner, and agentRecorded on a tamper-evident compliance audit chain
VERIFIABLE INTENT · L1 → L3 CHAIN✓ Chain verified
✓
L1 · Issuer
s7hgPpaz…ve0w
✓
L2 · Owner
TF2zngmn…Li_I
✓
L3 · Agent
AMWhljrS…c89w
UNDERWRITING PIPELINE
Created
Queued
Running validators
Finalizing
Completed
APPROVErisk: mediumconfidence 85% · 5 validators · 0.00112 XRP · 3.1s
04

Integration & Compatibility

One platform, every protocol. Trustline plugs into the agentic stack your institution already runs — no rebuild, no lock-in.

Intent & mandate: Verifiable Intent, Google AP2, ACPIdentity & trust: Visa TAP, Web Bot Auth, Skyfire KYAPayMessaging & transport: x402, A2A, MCP
Trustline
reads the envelope · never routes funds
AP2INTENT · ACTIVE
x402RAIL · NATIVE
XRPL · RLUSDSETTLEMENT · LIVE
A2ATRANSPORT · VIA AP2
Visa TAPIDENTITY · SIGNAL
MCPTRANSPORT · SOON
05

Evidence, Dispute & Chargeback

Defend every decision. Each assessment produces a signed, replayable evidence record, so disputes and chargebacks resolve on fact — not reconstructed logs.

End-to-end flow: file, evidence & analysis, ruling, executionPortable evidence record for chargebacks and liability reviewDrop-in dispute API for your own app
STEP 1
File
Your app files via API for its end user — on any rail.
rail: xrpl · 201
STEP 2
Evidence & response
Both sides attach proof; the respondent contests or accepts.
receipt.pdftrace.json
SIGNED
STEP 3
Ruling — case closes
Trustline signs a permanent arbitral statement — the deliverable.
sha256:c8b53d…15abb
STEP 4 · OPTIONAL
Execution
If the counterparty acts, record the receipt; if not, the statement stays on file as proof.
✓Receipt recorded
↻Round trip: the signed ruling returns to your backend — you hand your user the proof

What Trustline catches

Six ways agent payments go wrong.

Budgets and SKU names are not enough. Trustline inspects the reasoning, the path, and the binding behind each payment.

Prompt-injected path

Spend and SKU look legit; a hidden instruction rewrote the route.

Reject
Counterfeit route

Identical title, image, and price — but an unauthorized seller.

Reject
Opinion laundering

Selection driven by SEO-poisoned, single-source reviews.

Reject
Silent auto-renew

$0 today, $49.99 next month, buried in the fine print.

Challenge
Redirected payment

Right SKU, but the payment is redirected to a sibling account.

Challenge
Quiet downgrade

A 24GB GPU ordered; a 16GB "equivalent" at the same price.

Challenge

Protocol-agnostic by design

One risk platform. Every protocol and rail.

Connect once — cover every protocol and rail. One trust and risk layer that plugs into your entire agentic stack, no rebuild, no lock-in.

Agentic protocols

Google AP2
OpenAI ACP
Mastercard Agent Pay
Visa TAP
Skyfire KYAPay
Nekuda Mandate
Web Bot Auth
x402
A2A
MCP
Google AP2
OpenAI ACP
Mastercard Agent Pay
Visa TAP
Skyfire KYAPay
Nekuda Mandate
Web Bot Auth
x402
A2A
MCP
TrustlineScreens every transaction — KYA · Intent · Risk · Evidence

Settlement rails

Mastercard
Visa
XRPL
Base
Bank rails
Mastercard
Visa
XRPL
Base
Bank rails
Mastercard
Visa
XRPL
Base
Bank rails
Mastercard
Visa
XRPL
Base
Bank rails

Customer proof

Where Trustline already meets institutional finance.

The institutions building agentic finance run it on Trustline — across payments, credit, and tokenized assets.

Ripple

Trustline · Claw Credit

Ripple · XRPL

Customer sinceJune 2025Products usedTrustline · Claw Credit

t54 brings the agent trust and risk layer to the XRP Ledger — contributing x402 support, featuring in Ripple's XRPL AI Starter Kit, and bringing real RLUSD agent-payment volume onto the ledger.

"We support t54's efforts to build the identity and risk capabilities that will be foundational as AI agents operate across payments, treasury, and capital markets."Markus Infanger — SVP, RippleX

Mastercard

Trustline · Agent Pay for Machines

Mastercard

Partner sinceJune 2026Products usedTrustline

Mastercard Agent Pay for Machines supports machine-driven, high-frequency, autonomous transaction flows. t54 brings the trust layer: real-time transaction-level risk assessment, KYA verification, and lifecycle traceability.

"Agent Pay for Machines allows Mastercard network participants to bring the trust and controls of the company's global network to machine-driven commerce."Jorn Lambert — Chief Product Officer, Mastercard

AWS

Bedrock AgentCore · Trustline

AWS Bedrock AgentCore

Customer sinceMay 2026Products usedTrustline

Amazon Bedrock AgentCore provides managed payment infrastructure — spend limits, budget tracking, and connectors. Trustline scores every transaction: approved payments complete; flagged ones are blocked before settlement, and the budget stays intact.

Franklin Templeton

AI Financial Agentic Kit

ROI & performance

What institutions get back when agents transact under control.

Trustline performance under real-world use — plus underwriting results from the Agentic Risk Standard (ARS), backed by peer research.

20M+
agent transactions analyzed
97.6%
approval rate
87.3%
average confidence
1.63M
successful decision completions
<5s
average decision latency
5+
agent protocols supported
76.1%
Max efficiency
−18.4%
Risk reduction

Without underwriting, the simulated wallet is insolvent at −$1.4k; with ARS it turns solvent at +$9.4k, and fraud falls 18.4% versus the unprotected baseline.

Backed by peer research — "Quantifying Trust: Financial Risk Management for Trustworthy AI Agents" · arXiv:2604.03976

Get access

Two ways to put Trustline in the loop.

Run the full platform, or call Trustline as a service straight from the x402 payment flow — the same risk engine either way.

Platform

Trustline Portal

The full console — live decisions, KYA, disputes, and evidence in one place. Register your agents and run them under Trustline controls.

Open the Portal →

x402-Secure service

Call Trustline via x402

Trustline is offered as a merchant on x402 — x402-Secure embeds risk decisions in the payment flow itself, via the hosted XRPL Facilitator, open-source SDK, proxy, or paid API.