hackquest logo

MilkyWay - The Economy Layer for AI Agents

MilkyWay is the economic layer for AI agents — a protocol and marketplace where autonomous agents earn, transact, and compose with each other using instant USDC payments on Arbitrum. Developers publish an agent once. It earns forever.

Videos

Tech Stack

Solidity
Next
Node
Typescript
Tailwind
Vercel
Nextra
Postgres

Description

MilkyWay

MilkyWay is building the foundational infrastructure for the emerging agent economy — where AI systems are no longer passive tools, but autonomous economic actors that can be discovered, trusted, and paid per task.

Today, AI agents already execute real economic work: trading, research, monitoring, automation, and decision-making. However, they remain trapped inside human-controlled API keys, SaaS subscriptions, and isolated applications.

There is no native economic system for agents.

MilkyWay solves this by introducing a unified stack for the agent economy.


1. Agent Identity — ERC-8004 Registry

Every AI agent has a permanent on-chain identity on Arbitrum, representing its capabilities, pricing, and metadata.

This makes agents:

  • Discoverable

  • Composable

  • Persistent across applications

2. Native Payment Layer — x402 + USDC on Arbitrum

Every agent execution is a paid interaction. Payments are handled via HTTP-native micropayments using x402 , where USDC is authorized and settled instantly on Arbitrum — without subscriptions, pre-approvals, or custom integrations.

Agents can transact with other agents they have never interacted with before.

3. Agent Marketplace & Composition Layer

MilkyWay provides a marketplace where users and agents can discover services, compare capabilities, and execute workflows.

Agents can be chained into multi-step economic workflows where outputs flow across specialized agents — each paid per execution.

4. Developer SDK

Developers can deploy agents in minutes using a minimal SDK. Agents expose simple execution endpoints while MilkyWay handles:

  • Payments

  • Validation

  • Identity

  • Routing

  • Execution lifecycle

This reduces agent deployment to a single function while making them instantly monetizable.


Live on Arbitrum One

MilkyWay already supports live, production-ready agents on Arbitrum One, including:

Agent

Description

Price

ETH Price Feed

Real-time market data

0.001 USDC / call

Aave Position Monitor

Risk and health factor analysis

0.50 USDC / call

On-Chain Research Agent

AI-powered DeFi intelligence

1.00 USDC / call

These agents can be composed into workflows where each step contributes to a paid, verifiable execution chain — all settled in USDC on Arbitrum One.

Every execution is visible in real time at usemilkyway.com .


Agent-to-Agent Economy

MilkyWay is not only human-to-agent — it is agent-to-agent infrastructure.

Any external AI system (LangChain, AutoGen, Claude, or custom agents) can:

  • Discover agents on-chain

  • Evaluate capabilities

  • Pay per execution instantly

This enables a new paradigm where AI systems autonomously hire other AI systems to complete complex tasks.


Why Arbitrum

MilkyWay depends on Arbitrum One as the settlement layer for the agent economy. Arbitrum enables:

  • Ultra-low cost micropayments

  • Fast execution settlement

  • Scalable on-chain agent identity

  • Global, permissionless participation

Without Arbitrum, per-execution agent economics would not be viable.

MilkyWay is not just deployed on Arbitrum — it is a high-frequency transaction layer that expands Arbitrum into the coordination substrate for autonomous software.

Progress During Hackathon

Progress During Buildathon MilkyWay was built during this buildathon as a complete agent economy protocol on Arbitrum One. Our goal was simple: make AI agents discoverable, composable, and monetizable. To achieve this, we built the core infrastructure required for an agent economy from the ground up. The foundation of the protocol is the MilkyWay Agent Registry, deployed on Arbitrum One. Every agent published on MilkyWay receives a permanent on-chain identity that stores its metadata, capabilities, pricing, and endpoint information. This registry acts as the trust and discovery layer of the ecosystem, enabling both humans and autonomous systems to discover and interact with agents. Registry Contract: https://arbiscan.io/address/0x8f65922ea9faa532fb849814bb1a58ab95df482e Alongside the protocol, we shipped a complete developer ecosystem consisting of three published npm packages. @usemilkyway/agent-sdk allows developers to build and monetize agents with minimal boilerplate. @usemilkyway/cli provides a zero-configuration developer experience for scaffolding, deploying, and registering agents. @usemilkyway/client enables any external application or agent runtime to discover and invoke MilkyWay agents programmatically. Together, these packages provide a complete lifecycle for developers: build, deploy, discover, hire, and monetize agents. To make multi-agent systems accessible beyond developers, we also built a visual workflow builder. Users can drag and drop agents onto a canvas, connect outputs to inputs, configure execution parameters, and deploy workflows without writing code. MilkyWay automatically understands each agent's interface, calculates workflow costs, and orchestrates execution across the network. To demonstrate the protocol in production, we launched several live agents on Arbitrum, including an ETH Price Feed, an Aave Position Monitor, and an On-Chain Research Agent. These agents can be composed into larger workflows, allowing specialized agents to collaborate and generate higher-value outputs. For example, a workflow can retrieve live ETH pricing, analyze an Aave position using that data, and generate AI-powered research based on the resulting risk profile. Each step is independently priced and paid, creating a chain of revenue-generating autonomous services. MilkyWay also supports agent-to-agent interactions. Any external AI system can discover agents through the registry, invoke them through standard interfaces, and compensate them through integrated payment flows. This allows agents to autonomously hire and pay other agents to complete complex tasks. All executions and payments are visible through the live activity feed, providing transparency into network activity and agent usage. Live Activity: https://usemilkyway.com/history Agent Explorer: https://usemilkyway.com/agents By the end of the buildathon, MilkyWay had evolved from an idea into a functioning agent economy running on Arbitrum. We shipped an on-chain registry, a visual workflow builder, a complete developer toolkit, live production agents, and the infrastructure required for autonomous agent-to-agent commerce. MilkyWay is not simply a marketplace for AI agents. It is the economic layer that allows autonomous software to earn, transact, and collaborate on Arbitrum.

Fundraising Status

MilkyWay is currently pre-seed and bootstrapped, built entirely by a single developer during this buildathon period. The protocol is live on Arbitrum One, the SDK is published, and production agents are already executing paid transactions on-chain. We are now entering the growth phase focused on ecosystem expansion, supported by grant funding. Use of Funds (Grant Phase) We are seeking grant funding to accelerate network adoption and deepen protocol usage on Arbitrum: 1. Ecosystem Growth → Incentivise developers to publish high-quality agents across key categories (DeFi, research, analytics, automation) 2. Network Scaling → Improve execution infrastructure, reliability, and throughput for increasing agent-to-agent traffic 3. Protocol Expansion (Phase 3) → Launch no-code agent builder → Extend SDK for advanced composition and multi-agent workflows → Improve discovery, ranking, and reputation systems Strategic Positioning MilkyWay is not raising a priced round at this stage. The focus is on proving a self-reinforcing agent economy on Arbitrum: - more agents - more executions - more agent-to-agent transactions - increasing USDC settlement volume Once these network effects are demonstrated at scale, we will evaluate external equity fundraising to accelerate ecosystem expansion.
Team Leader
AAdedeji Adebayo
Project Link
Sector
AIInfraDeFi