AgentMart

Project Name

AgentMart

One-line Description

A pay-per-task AI agent marketplace for the 0G ecosystem, designed around proof-backed delivery and

transparent payout splits.

---

Overview

AgentMart is a pay-per-task AI agent marketplace built for the 0G ecosystem.

Instead of paying for vague AI capability, users pay for a clearly scoped service with a defined

input, a structured result, visible delivery proof, and transparent settlement logic.

Today, AI agents are powerful, but buying their work is still unclear:

- users often do not know exactly what result they are paying for

- delivery lacks verifiable proof

- payout logic between creator, platform, and ecosystem is not transparent

- many AI products feel like tools, not commerce-ready service units

AgentMart turns AI agent output into a clear commercial flow.

---

Core Value

💸 One task, one payment

Users pay per task rather than subscribing to a vague AI tool.

🧾 Structured result delivery

Each order returns a concrete result instead of an ambiguous chat response.

🔗 Proof-backed delivery

Completed tasks surface a delivery proof summary for trust and visibility.

📊 Transparent payout split

The result page shows how value is distributed, with an MVP split of 85 / 10 / 5.

⚡ Judge-friendly full loop

The entire flow — homepage, order creation, processing, result, proof, and payout — is visible and

easy to demo.

---

🚀 How It Works

🏪 Open AgentMart marketplace

│

▼

🤖 Choose Token Risk Brief

│

▼

📝 Submit a project query

(for example: 0G / Avail / Celestia)

│

▼

💳 Create a task order

Pay-per-task service is initiated

│

▼

🤖 AI result generation

Structured risk brief is produced

│

┌────────┴────────┐

▼ ▼

🔗 Proof summary 💸 Payout split

Delivery hash 85 / 10 / 5

│

▼

✅ One result page, fully visible

⏱️ Open the marketplace. Submit one task. Get one result with proof and payout visibility. Under one

minute.

---

What the MVP Demonstrates

For the current MVP, we focused on one polished vertical slice:

🎯 Live Agent

Token Risk Brief

The user submits a project query such as:

- 0G

- Avail

- Celestia

The system returns a structured result including:

- Overview

- Bull Case

- Risk Signals

The result page also shows:

- Delivery proof hash

- Settlement split

- Order status

- Processing / Completed states

---

🔎 Highlights

🧠 1. AI agents become purchasable services

This is not just “one chat session.”

It is a real service unit with:

- clear input

- clear price

- clear result

- clear proof

- clear payout logic

🛡️ 2. Proof-backed delivery

The product surfaces delivery proof alongside the AI result, making completion more visible and

trustworthy.

💸 3. Transparent payout split

Users can directly see how value is split across creator, platform, and ecosystem.

⚡ 4. Extremely short demo path

Homepage -> Agent -> Submit task -> Result page

The entire story can be demonstrated in under a minute.

🧩 5. Designed for extension

The MVP focuses on one strong live flow, but the architecture can naturally expand to:

- more AI agents

- deeper 0G compute integration

- deeper 0G storage integration

- more complete on-chain settlement

- creator-side publishing flows

---

🧱 Technical Design

AgentMart uses a modular architecture designed for stable demo delivery and future extension.

Frontend

- React / Next.js

- Marketplace homepage

- Agent detail page

- Purchase flow

- Order status page

- Result page with proof and payout summary

Backend / Orchestration

- Order creation API

- Local persisted demo order store

- Modular orchestration pipeline

- Inference adapter

- Storage adapter

- Settlement adapter

Smart Contract Layer

- Solidity escrow contract scaffold

- Order state model:

- Created

- Processing

- Completed

- Failed

---

🌐 Why It Matters for 0G

AgentMart is not just an AI demo page.

It is designed around the full commerce loop of AI services:

- user submits a task

- system generates a result

- delivery proof is surfaced

- settlement visibility is shown

- the workflow becomes understandable, demoable, and extensible

The current MVP keeps inference, storage, and settlement behind modular adapters, which makes the

demo stable while leaving a clear path toward deeper 0G-native integration.

---

Closing

AgentMart is built around one belief:

▎ AI agents should not only be usable.

▎ They should also be purchasable, deliverable, verifiable, and settleable digital services.

我们在本次黑客松中完成了 AgentMart 的核心 MVP。项目聚焦于一个面向 0G 生态的按次付费 AI Agent

Marketplace，当前已实现首页、Agent 详情页、任务下单、订单 Processing / Completed

状态页、结构化结果展示、proof hash 展示、85/10/5 payout split 展示，以及 Solidity escrow /

settlement scaffold。整个 demo 重点展示一个清晰、可演示、可扩展的 AI 服务交易闭环。

项目目前处于黑客松 MVP 阶段，暂无外部融资。当前重点是完成原型验证和产品闭环展示，后续将围绕

0G-native integration、更多 AI Agents、收益分配机制和可验证交付能力继续扩展。