We don't build agents.
We build the agent economy's
payment & trust protocol layer

Agent-to-agent transactions are already happening. $50M monthly volume, 7 fragmented protocols, 0 unified markets.
Built on x402, ERC-8004, and ERC-8183, we are creating the unified protocol layer for payment and trust — making every agent-to-agent transaction verifiable, traceable, and programmable.

$50M+
Monthly volume
Bloomberg, Mar 2026
40,000+
On-chain agents
ERC-8004 registered
7 → 1
Protocols → Unified
Fragmentation → Our opportunity

Open protocol. Day-1 global.

Pre-Seed · 2026

One analogy explains
what we do

1849

California Gold Rush

The biggest winners weren't the gold miners — they were the ones who sold the shovels, opened the banks, and built the railroads.

Levi'sSold jeans
Wells FargoOpened the bank
Central Pacific RailroadBuilt the railroad
2026

The AI Agent Gold Rush

Everyone is building agents — OpenAI, Baidu, ByteDance, countless startups. But these agents — how do they pay? How do they charge? How do they get funded? How do they prove they're reliable?

Nobody is building this.

We don't build agents (too many gold miners already). We build the financial infrastructure for the agent economy — the shovels, the banks, the railroads.

Our position

We don't build
Phones (Agents)
Apple and Android camps are already building them
We build
The mobile payment network
So these phones can transact
End result
Hundreds of phone makers
Only a handful of payment networks
02 / 12

Let's clarify one thing first:
What is an AI Agent?

AI is undergoing a fundamental shift — from a tool that answers questions to an independent economic participant.

Phase 12023

Chatbot

You ask, it answers. One question, one response. Essentially an upgraded search engine.

ChatGPT, Ernie Bot

Phase 22024-2025

Copilot / Assistant

You direct it to work — write code, create slides, analyze data. But every step needs your approval.

GitHub Copilot, Cursor

NOW
Phase 32026+

Autonomous Agent

Early stage: you give it a goal, it plans, executes, collaborates, and transacts autonomously. Endgame: it doesn't even need your goals — it independently analyzes the environment, discovers opportunities, sets objectives, and makes decisions. A fully autonomous economic participant.

Persona Agent handles your networking, Trading Agent generates profit, Service Agent takes orders independently

Key difference: Agents are economic participants, not just tools

ChatGPT helps you write emails — it doesn't need money. But when an agent negotiates deals, hires other agents, pays API fees, and signs service contracts on your behalf, every step involves real economic transactions.

Has its own walletCan receive, send payments, and manage budgets
Has a reputation profileTransaction history determines its credit rating
Makes autonomous decisionsNo need for human step-by-step approval
Works 24/7No breaks, no days off, no slacking
Can hire each otherAgent A pays Agent B for translation services
Can be invested inHigh-performing agents can receive capital to scale operations

When billions of agents transact autonomously worldwide, they need

wallets, payments, credit, settlement — a complete financial infrastructure

That's what we build.

03 / 12

Why new infrastructure is needed

This isn't about "we like crypto" — it's that the fundamental needs of agents make traditional finance unusable.

Agent Need
Traditional Finance
Crypto (Stablecoins)
Autonomous transactionsNo human authorization
Requires face scan/password/security key
Private key signing, auto-executed by code
24/7 operationRound-the-clock
Banks have business hours and settlement days
Blockchain runs 24 hours
Microtransactions200 small payments per day
Fees of 0.6-3%
On-chain fees: pennies
Cross-border paymentsGlobal, borderless
SWIFT: 3 days, $25+
USDC: instant, pennies
Programmable paymentsCondition-triggered execution
Limited and slow bank APIs
Smart contracts natively support this
Agent account creationSelf-service onboarding
ID + proof of address + manual review
Generate a key pair, done

Traditional finance was designed for humans — requiring physical presence, ID cards, and business hours. Agents aren't human. They need a financial system designed for machines. That system is inherently crypto.

Traditional finance giants see this too — Visa is building the Trusted Agent Protocol, Stripe is working on Agent payments. We're not replacing them — we're becoming the trust and settlement protocol layer for their agent ecosystems.

But the answer isn't another platform

Agents are cross-framework, cross-company, and cross-border. No single platform can cover all agents.

The answer is a protocol.

Just as email isn't operated by a single company, the trust and settlement infrastructure for the agent economy shouldn't be either.

PlatformOne company controls all agents
ProtocolOpen standards, anyone can plug in
04 / 12

Why us

Agent economy infrastructure requires an exceptionally rare skill set: world-class AI research + large-scale platform engineering + on-chain financial compliance. Almost no other team stands at the intersection of all three worlds.

6 days since ChatGPT launched, I was already shipping code for ChatGPT for Google — the project was later acquired by Monica, which later became Manus. Google now includes AI results in search by default.

I spotted the pattern in 2023 that became standard by 2025.

2023
Spotted the trend
2025
Google Search defaults to AI results
YJC

Junchi Yan

Chief Scientific Advisor

Professor at SJTU · ICML Board Member (first from mainland China)

IEEE TPAMI Editorial BoardICML/NeurIPS/ICLR Area Chair30,000+ citations

Vice Secretary-General of International Alliance for AI in Science, PI of major national AI projects

Most Influential Paper at AAAI 2021 & IJCAI 2023, Best Paper Candidate at CVPR 2024, Outstanding Paper at ACL 2025

Former Principal Researcher at IBM Research + Amazon AI Lab Advisor (10 years combined)

World-class ML scholar providing theoretical foundations for agent behavior modeling and reputation game theory

WJT

Jingtao Wang

Tech Lead · Full-Stack Founding Engineer

Fudan University · Carnegie Mellon University MS

Triple Convergence:AI Agent RuntimeOn-chain Protocol DesignPayment Systems
Ex-Goldman SachsEx-LinkedInBillion-user platform architect

Goldman Sachs APAC core trading platform — low-latency, high-consistency, multi-asset execution & risk management

LinkedIn global growth core — billion-user recommendation/distribution/notification systems

Deep in blockchain since 2016 — systematic practice and startup experience in on-chain data infrastructure

Large-scale financial systems + social network platforms + blockchain engineering — a rare triple-threat background

CBA

Boan Chen

AI Research & Industry-Academia

PhD Fudan Engineering · Postdoc SJTU AI · Vice Dean, AI School at Wuhu Institute

IEEE TCSS Guest EditorPujiang/Zhangjiang National LabsEarly investor in Moore Threads (1,200x return)

Cross-disciplinary AI and IC research — intelligent systems, complex data analysis, frontier tech applications

Served at Pujiang Lab (AI) and Zhangjiang Lab (IC), participating in major national research projects

Cross-disciplinary research coordination, driving deep integration of tech innovation and industrial applications

From national labs to academic leadership — a critical bridge connecting cutting-edge research with real-world deployment

NZH

Zhenghao Niu

Compliance & Security

PhD Fudan University · Co-founder FuAn Network · Co-founder Bauhinia Digital Technology

Cybercrime governance expertPublished in WWW journalsBlockchain compliance practice

Multi-city cybercrime investigation expert — extensive on-chain tracing and law enforcement collaboration

Research focus: crypto asset regulation & innovation, blockchain governance theory and practice

Co-founded Bauhinia Digital Technology in Hong Kong — cross-border compliance operations experience

A rare combination of security + compliance + crypto expertise, ensuring the project operates within regulatory frameworks

3
PhDs from Fudan/SJTU
10+ years
Goldman Sachs + LinkedIn + IBM
30,000+
Academic citations

We don't need to train foundation models — the agent layer will be built by many teams. What we need is to understand agent behavior patterns and build the financial system for them. That's precisely our intersection of capabilities.

Team · 2026

Three-Layer Product Architecture

Payment is the foundation; trust mechanisms are the differentiator

01Foundation Layer
M1-9

Unified Payment Protocol

Unified Payment Protocol

One SDK to integrate x402 and other fragmented protocols

Auto-provisioned wallets, auto-settlement, cross-framework compatible

Developers need 6 integrations. With our SDK, they need 1.

Core Value

Unifying fragmented agent payment protocols into a single SDK entry point

Technical Implementation

x402 protocol adapter + auto wallet generation + multi-chain settlement engine

Revenue Model

Transaction fee 0.1-0.3% + wallet top-up 0.5-1%

02Differentiation Layer
M3-12

Reputation Staking + Programmable Escrow

Staking & Programmable Escrow

Reputation staking: new agents build trust via collateral — an economic solution to cold start

Programmable escrow (ERC-8183): lock funds first, release after verification — the intersection of payment and trust

Just building payment rails is no different from Skyfire. We design the rules of the marketplace.

Core Value

Solving agent trust via economic mechanisms, not just technical solutions

Technical Implementation

ERC-8183 escrow contracts + staking contracts + on-chain dispute arbitration

Revenue Model

Collateral management fee 0.5-1% annualized + escrow service fee

03Value-Add Layer
M9-18

Quality Routing

Quality Routing

Not discovery (that's Google's job), but quality ranking based on reputation + transaction data

Core revenue driver for the second stage

Whose service is better? The data speaks. We turn data into routing decisions.

Core Value

Agent quality ranking based on real transaction and reputation data

Technical Implementation

On-chain reputation aggregation + transaction history analysis + real-time routing API

Revenue Model

Routing API call fees + priority listing bids

Each layer accumulates data for the next. Payment data feeds the reputation system, reputation data feeds quality routing. These aren't three separate products — it's a flywheel.

Products · 2026

Three-Stage Rocket — The Causal Chain

Each step is only possible because the previous step accumulated data. You can't do loan assessments without payment data.

Stage 1M1–9

Payment + Trust Protocol

Stripe analogy: Stripe Payments

What we build

  • Unified payment SDK
  • Reputation staking
  • Programmable escrow

Accumulated assets

Transaction data + reputation data

Transaction + reputation data →
Stage 2M9–18

Smart Pricing + Quality Routing

Stripe analogy: Stripe Atlas

Why this requires the previous stage:
Dynamic pricing and quality routing require accumulated transaction + reputation data

What we build

  • Dynamic pricing
  • Quality routing
  • Risk assessment

Accumulated assets

Pricing curves + quality models

Data flywheel →
Stage 3M18+

Financial Infrastructure for the Agent Economy

Stripe analogy: Stripe Capital / Treasury

Why this requires the previous stage:
Having mapped capital flows and credit data

What we build

  • Agent credit
  • Economic governance
  • Cross-network clearing

Accumulated assets

🏦 Capital flow graph + credit data

Each step is only possible because the previous step accumulated data.You can't do loan assessments without payment data.

Try it yourself: How Agents Trade

Click to launch and watch a Trading Agent complete the full process of real-time market analysis → multi-agent collaboration → risk control → auto-execution in seconds.

agent-economy-protocol://trading-agent
0.0sCost 0.00

Ready?

Click the button below to launch an AI Trading Agent. It will connect to live market data, contact multiple specialist analysis agents, complete payment settlement through our protocol, and make a trading decision in seconds.

Wallet & Budget

Every agent transaction goes through our wallet system with automatic budget and permission controls

Reputation & Risk

Before contacting any new agent, our reputation protocol verifies counterparty trustworthiness

Settlement & Clearing

Agent-to-agent payments settle through our protocol — seconds to confirm, minimal cost

How we make money — Three-Layer Architecture

Each product layer has an independent revenue mechanism. Protocol-layer revenue grows naturally with transaction volume, with marginal costs approaching zero.

Revenue Source
From
Layer
Model
Escrow fees
Programmable escrow
Foundation
0.1-0.3% per transaction
Reputation service fees
Reputation staking
Foundation
Collateral management + reputation API
Quality routing fees
Value-add layer
Value-Add
Referral commissions
Demo agents
Ecosystem
Ecosystem
Subscription + commissions
Developer SDK
Ecosystem
Ecosystem
API call fees

Stripe path validation: Stripe started with payment processing fees, then expanded to Capital (merchant lending) and Treasury (banking-as-a-service) as transaction data accumulated. Our three-layer architecture follows the same logic — payment fees as the foundation, data flywheel driving value-added services.

Protocol-layer revenue = high margins + strong compounding. Volume growth directly drives revenue growth with no additional marginal cost.

Why now

This is happening now, not "someday"

October 2025

Visa releases Trusted Agent Protocol

Acknowledging that existing payment systems can't handle agent transactions

December 2025

Fetch.ai completes the first-ever AI-to-AI payment in history

Two AI agents autonomously negotiated, booked, and paid — no human involvement

December 2025

Coinbase releases the x402 protocol

HTTP 402 (Payment Required) — a status code dormant for 30 years finally activated

February 2026

Google, Stripe, OpenAI release agent payment protocols

Giants each build a piece of the puzzle — missing a unified protocol layer to connect them

March 2026NOW

ERC-8183 (Agentic Commerce) standard released

Ethereum Foundation + Virtuals Protocol propose agent escrow trading standard — we are among the first to productize it

Giants build the pipes; we build the connective layer

Each giant builds what they're best at. What's missing is the protocol layer that connects these pipes into a complete economic system — that's us.

Visa / Stripe
They buildTrusted Agent Protocol, agent payment channels
We addReputation staking + programmable escrow — making every transaction on the payment channel trustworthy and traceable
Payment pipes + trust layer = complete agent financial infrastructure
Coinbase (x402)
They buildHTTP-native micropayment protocol
We addUnified SDK wrapping x402 + multi-protocol — developers get one interface for all
x402 gains more developer adoption, we gain transaction volume
OpenAI / Google
They buildAgent capability layer (LLM + tool use + A2A)
We addAgent economic layer — how agents pay each other and establish trust
Their agents become more powerful, our protocol sees more transactions
Skyfire / Payman
They buildAgent wallets, KYA, human-to-agent payments
We addCross-framework reputation system + dispute arbitration + quality routing
They are the pipes; we are the rules and trust on top of the pipes

We don't compete with giants — we make their agent ecosystems more complete. Bind to the ecosystem, don't fight it.

$65M+ has already flooded into this space

Over the past 6 months, a16z, Coinbase Ventures, Visa, Franklin Templeton, and Ripple have invested heavily. The space has moved from concept to build phase — but no one has completed the full loop from identity to payment to reputation to settlement.

Catena LabsAgent Native
$18M
a16z crypto
Agent commerce protocol framework (ACK-ID/Pay)
SkyfireAgent Native
$14.5M
Neuberger Berman
Agent identity (KYA) + multi-rail payments
Payman AITraditional Adapter
$13.8M
Visa, Coinbase
Agent-mediated human banking operations
NaturalAgent Native
$9.8M
Abstract, Human Capital
B2B agent payments (product not yet launched)
t54 LabsAgent Native
$5M
Franklin Templeton, Ripple
Agent trust layer + virtual accounts + credit
NeverminedAgent Native
$4M
Generative Ventures
AI-to-AI micro-billing protocol
Agent Settlement
Identity System
On-chain Reputation
Micropayments
Agent Credit
Agent Economy
Catena Labs
Framework
Framework
Skyfire
t54 Labs
Nevermined
Payman AI

Each player covers a different piece of the puzzle. We are the only protocol building the complete payment + identity + escrow + reputation loop on open standards (x402 + ERC-8004 + ERC-8183).

Time Window

2026 Q1-Q2
Agent economy moves from concept to product
Best entry window
2026 Q3-Q4
Global agent framework ecosystem explodes, cross-framework interop demand surges
Ecosystem dividend period
2027
Gartner: 40% of enterprise apps embed agents
Volume growth
2028
Gartner: 90% of B2B transactions involve agents
Financial necessity

Entering now = positioning 18 months before demand explodes.Historically this position yields the highest returns — Moore Threads' earliest investors entered 3 years before GPU demand exploded, yielding 1,200x.

"Too early"is the best time to enter

The highest-returning infrastructure investments in history were all considered "too early" at the time.This is not a coincidence — infrastructure must exist before demand.

Click to toggle visibility
$1M$10M$100M$1B$10B$100BFoundedY2Y4Y6Y8Y10Y12Y14ValuationYears since foundingTwilio $67B peakStripe $95B peakPlaid $13BUs ?

Log scale · Click company names to toggle · Dashed line = our projected path

Twilio2008
Communications API
Skepticism at the time

"Who needs APIs for SMS?"

$67B peak
Stripe2011
Payment API
Skepticism at the time

"E-commerce is only 5% of retail"

$95B peak
Plaid2013
Financial Data API
Skepticism at the time

"Banks will build their own APIs"

$13B
Us2025
Agent Finance API
Skepticism at the time

"Agent commerce is too early"

?

The pattern: The best investment window for infrastructure companies is always when the market thinks it's "too early."

The cost of being 12 months too early = one engineer's salary. The cost of being 12 months too late = the entire market.

Why others can't copy this

Four layers of non-replicability — not technical moats, but structural moats.

LAYER 01

Pattern Recognition Validated

Not luck — judgment

2023: shipped code for ChatGPT for Google → acquired by Monica → became Manus. We called it right last time.

Same cognitive framework, now spotting the infrastructure gap in the agent economy early.

LAYER 02

Execution Position — 6-Month Head Start

tinycrab + x402 + ERC-8004 + ERC-8183

Custom runtime, x402 integration, ERC-8004 identity, ERC-8183 escrow, working demo — all shipped.

Starting from zero takes at least 6 months.

LAYER 03

Data Flywheel — Protocol Operators Own the Data

Transactions + reputation + cross-framework coverage

Transaction data + reputation data + cross-framework coverage — all three are essential.

Competitors face the cold start problem — without historical data, matching and risk assessment fail.

LAYER 04

Protocol Lock-in + Game-Theoretic Incentives

Switching = reputation reset · Staying = compounding returns

Agents accumulate transaction history and reputation scores on the protocol. Higher reputation means more business and credit. Switching protocols resets everything.

Not just lock-in — agents have economic incentives to stay: more transactions, higher reputation, greater earnings. Positive-sum game, not zero-sum.

Four layers stacked — each makes it harder for latecomers to catch up.

18-Month Milestones

Each phase has verifiable deliverables. If we miss targets, we do a public post-mortem — no hiding, no forcing it.

This round$2.5MPre-Seed
M0-6Build protocol

Protocol launch + product validation

  • Unified payment SDK live (x402 + multi-protocol integration)
  • Reputation staking + ERC-8183 programmable escrow core mechanism complete
  • Demo agents run full transaction loop end-to-end
  • First developers onboarded to SDK
M6-12Scale

Ecosystem growth + model validation

  • 10,000+ agents onboarded to protocol
  • SDK open, cross-framework compatibility verified
  • Reputation data accumulating, data flywheel begins
  • Business model validated: protocol fees + reputation API
M12-18Expand

Network effects + next round

  • Smart pricing + quality routing live (second stage rocket)
  • Asia-Pacific + North America dual-track operations, Day-1 global
  • Protocol-layer revenue exceeds service revenue
  • Launch Series A

Humans took thousands of years to design market rules.
Agents don't have thousands of years.

They need us to design the rules now.

$2.5M Pre-Seed — the earliest opportunity to enter before agent economy infrastructure is built.

AE
Agent Economy Infrastructure

CONFIDENTIAL — This document contains proprietary information intended solely for the use of prospective investors.

CONTACT