SmartPay – Project Overview

Inspiration

The gig economy is changing fast. Today, when we need expert help, we usually turn to human freelancers on platforms like Upwork or Fiverr.

But we believe the next workforce won’t be limited to humans alone — it will include autonomous AI agents that can think, reason, and execute tasks instantly.

While AI capabilities have advanced rapidly, payments have not. Traditional payment rails and even most crypto systems are too slow, expensive, and inefficient for a future built on instant, micro-transactional, machine-to-machine work.

We asked ourselves a simple question:

What if you could instantly hire a specialized AI expert and pay them seamlessly on-chain?

That question became the foundation of SmartPay.

What We Built

SmartPay is a decentralized marketplace for specialized AI agents. Instead of hiring general-purpose AI tools or human freelancers, users can browse and hire AI agents built for specific professional tasks.

Each agent has a clear role:

Vault – Financial analysis, market insights, and risk evaluation
Forge – Development tasks like code generation, debugging, and optimization
Aura – Creative work such as branding, copywriting, and storytelling
Compass – Product strategy, roadmap planning, and user-centric decision making

Users unlock an agent by making a one-time on-chain payment using MNEE stablecoin. Once the payment is confirmed, the agent activates instantly and delivers results within seconds.

How We Built It

We designed AgentPay around speed, reliability, and low latency.

AI Engine

Agents are powered by Google Gemini 2.5, with careful model selection:

Gemini 2.5 Flash – Fast, responsive tasks like finance and creative work
Gemini 2.5 Pro – Deep reasoning tasks like development and product strategy

Payments

We integrated MNEE stablecoin as the exclusive payment method, enabling instant settlement and a true pay-and-use experience.

Frontend

Built using React and TailwindCSS, designed with a clean, dark-mode Web3 aesthetic.

Backend

A lightweight serverless architecture listens for on-chain payment confirmations and securely triggers the corresponding AI agent only after payment is finalized.

Challenges We Faced

One of the biggest challenges was orchestrating the payment-to-action flow.

We had to ensure:

$$ \text{AI Execution} \Longrightarrow \text{Only After On-Chain Confirmation} $$

At the same time, users expect AI responses to feel instant. Balancing blockchain confirmation times with real-time AI execution required careful tuning of listeners, retries, and UI feedback states.

Another challenge was prompt engineering — making sure each agent consistently behaved like a true professional in its domain rather than a generic chatbot.

What We Learned

Model selection matters — matching the right AI model to the right task dramatically improves results.
Stablecoins are essential for machine-to-machine commerce due to predictable pricing and fast settlement.
Role-defined AI agents feel more “alive”, task-focused, and payment-gated.
On-chain payments + AI unlock new product categories that weren’t possible before.