Across India, developers are experimenting with a new question in AI and Web3: What happens when software can pay for the services it needs?
AI agents are becoming increasingly capable of taking actions on behalf of users. They can search for information, call APIs, analyze data, generate code, plan trips, and coordinate multiple services.
But there is still a missing piece: how does an agent pay for the things it needs? This is where x402 comes in.
x402 is a payment protocol designed for internet-native transactions. It allows services to request payment as part of an API or HTTP interaction, creating a way for software to pay for software without relying on traditional accounts, subscriptions, or manual checkout flows.
Recent Algorand hackathons gave Indian builders an opportunity to experiment with this idea through working prototypes. Hundreds of developers participated across events, building everything from AI-powered recruitment and autonomous travel to research platforms, disaster response, coding services, and financial controls for AI agents.
What makes these projects interesting is not simply that they use blockchain payments.
They show different ways payments could become part of an AI agent's workflow.
Here are some of the projects that caught our attention.
1. Agents are becoming customers
One of the clearest patterns across these builds is the idea that an AI agent can become a customer of another service.
Instead of a person choosing a service, paying for it, and using it, an agent can decide that it needs a particular capability, make the payment, and continue its task.
🥇 BlockHacker
AI-powered recruitment with pay-per-use resume analysis
Built at NexVerse, BlockHacker created an AI-powered Applicant Tracking System where resume analysis becomes a paid API service.
A user pays $0.50 USDC through x402, and the system verifies the on-chain payment before running its AI analysis through Groq. Each request can be processed independently, creating a simple pay-per-use model.
The interesting part is the pricing model. Instead of requiring a subscription for an occasional task, an AI capability can become a service that users pay for exactly when they need it.
DoraHacks: https://dorahacks.io/buidl/47564
🥈 AgentPay by Argo Operators
Giving AI agents a way to pay for AI services
Argo Operators built AgentPay, an x402-powered AI payment router that allows agents to pay for AI requests in USDC.
The system can route prompts to AI providers while tracking settlements, budgets, and provider performance.
As AI agents become better at choosing which tools and services to use, payments become part of the workflow. AgentPay explores the infrastructure needed to make those transactions possible.
DoraHacks: https://dorahacks.io/buidl/47552
🥈 VoyaAI
An autonomous travel agent that can pay for the services it uses
VoyaAI uses Algorand x402 micropayments to purchase real-time travel services, build itineraries, and dynamically replan trips.
Travel is a natural example of agentic commerce. An AI agent planning a trip may need multiple services along the way, from information and recommendations to travel-related services.
VoyaAI explores what that experience could look like when the agent can handle those transactions itself.
DoraHacks: https://dorahacks.io/buidl/47553
🥉 CareerVolt
Six AI agents working across the career journey
CareerVolt brings together six specialized AI agents covering job matching, resume analysis, career roadmaps, job-market insights, interview preparation, and an AI copilot.
The platform uses Algorand x402 to offer these capabilities as pay-per-request services.
Rather than treating AI as one large bundled product, CareerVolt explores a model where individual capabilities can become independently accessible and monetizable services.
DoraHacks: https://dorahacks.io/buidl/47554
2. Agents need to buy information
The opportunity goes beyond AI assistants buying consumer services.
Agents also need access to information.
Datasets, APIs, research tools, real-time information, and specialized software can all become resources that an agent needs to complete a task.
🥇 Research Swarm
An AI research platform that can pay for the resources it needs
Built at CodeRush 2.0, Research Swarm takes a research project from an initial idea through literature review, data collection, analysis, drafting, citations, and refinement.
Its agents can use Algorand x402 to access paid research tools, datasets, APIs, and academic resources.
This creates an interesting machine-to-machine payment model.
When an agent needs a particular resource, it can access the service and handle the associated payment as part of the research workflow.
The payment is no longer a separate step performed by a human. It becomes part of the task itself.
DoraHacks: https://dorahacks.io/buidl/48018
🥈 Suraksha Setu
Helping disaster-response agents access real-time data
Suraksha Setu combines disaster simulation, resource tracking, route planning, and dynamic rerouting into a disaster-response platform.
Its agents can use Algorand x402 to access and pay for external real-time data.
That means the system can bring additional information into its decision-making process while keeping the payment flow automated.
The project highlights a practical use case for machine-to-machine payments. When software needs information from an external service, access and payment can happen programmatically.
DoraHacks: https://dorahacks.io/buidl/47620
🥉 Yukti: Intelligence That Sharpens Code
A coding agent that can pay for other AI services
Yukti is a model-independent coding harness that uses AST and CFG-based repository intelligence to identify relevant code context.
It then verifies and self-corrects generated code inside a sandbox. Its agentic endpoint uses Algorand x402 for paid AI-to-AI access to coding services.
This points toward another possible model for developer tools.
A coding agent doesn't have to perform every task itself. It can call another specialized service, pay for that capability, receive the result, and continue its workflow.
In this model, AI systems can become both service providers and customers.
DoraHacks: https://dorahacks.io/buidl/48019
3. If agents can spend, they need eules
Giving agents the ability to pay introduces another problem.
How much should an agent be allowed to spend?
And perhaps more importantly:
What should an agent be allowed to buy?
As agents move from making recommendations to taking actions, developers will need ways to define budgets, permissions, transaction limits, approved services, and human approval requirements.
Several builders are already exploring this layer.
🥇 ContractLens
Contract analysis as a pay-per-use AI API
ContractLens lets users upload a contract, pay $0.50 USDC on Algorand TestNet through x402, and receive a structured AI-generated risk report.
Each request is verified on-chain before the analysis is delivered, with no signup or subscription required.
It demonstrates a straightforward pay-per-use model for an AI service. A user has a specific need, pays for that request, and receives the result.
DoraHacks: https://dorahacks.io/buidl/47565
🥈 MandateGuard
Giving autonomous agents rules for how they spend
MandateGuard focuses on the financial control problem around autonomous agents.
The platform validates payment requests against budgets, transaction limits, approval rules, and spending restrictions before execution. x402 and Algorand provide the payment flow and an auditable transaction record.
This becomes increasingly important as agents gain the ability to spend on behalf of users or organizations.
The question is no longer only whether an agent can make a payment.
It is whether that payment is allowed.
MandateGuard puts those controls directly into the payment workflow.
DoraHacks: https://dorahacks.io/buidl/47574
🥉 Agent Spend Policy Engine
A programmable spending boundary for AI agents
Agent Spend Policy Engine lets administrators define transaction limits, daily budgets, approved services, and approval requirements for autonomous agents.
Payment requests are checked before execution. Approved payments can settle through x402 on Algorand, while suspicious requests can be blocked or routed for human approval.
It explores an important piece of infrastructure for an agentic economy: giving software the ability to spend while keeping humans in control of the boundaries.
DoraHacks: https://dorahacks.io/buidl/47562
Three patterns emerging from these builds
Looking across these projects, a broader picture starts to emerge.
1. AI capabilities can become paid services
BlockHacker, ContractLens, and CareerVolt show how individual AI capabilities can be exposed as services and monetized on a pay-per-use basis.
Instead of putting everything behind a monthly subscription, a service can charge for a specific action.
2. Agents can become buyers
VoyaAI and Research Swarm explore a different shift.
The customer doesn't always have to be a person.
An AI agent can need a dataset, API, research tool, travel service, or specialized AI capability. If that service has a price, the agent needs a way to pay for it.
This creates the possibility of machine-to-machine commerce, where software can discover, purchase, and consume services as part of completing a task.
3. Autonomous spending needs guardrails
MandateGuard and Agent Spend Policy Engine highlight the other side of the equation.
If agents can spend money, developers need to define what they can spend, how much they can spend, which services they can access, and when a human needs to approve a transaction.
Payment infrastructure and policy infrastructure therefore become closely connected.
The bigger idea: an economic layer for AI
The most interesting thing about these projects is not any single application.
It's the design space they open up.
An AI agent doesn't have to be limited to generating an answer.
It can:
-
-
-
access a dataset
-
call an API
-
purchase a specialized capability
-
retrieve real-time information
-
use another AI model
-
pay for compute or infrastructure
-
coordinate with other services
-
-
And each of those interactions can have an economic layer.
This changes the role of payments.
Today, payments are usually something that happens after a human makes a decision.
In an agentic system, payment can become part of the decision-making loop:
Need a service → discover the service → make a payment → receive the result → continue the task
That's the bigger opportunity behind x402.
Not simply making payments programmable, but making economic interactions programmable.
The question for builders is no longer just: What can an AI agent do?
It's also: What would an AI agent be willing to pay for?
Take your build further
The projects coming out of these hackathons are early experiments, but they point toward a much larger opportunity.
If you built one of these projects, the next step is turning the prototype into something people can actually use and pay for.
If you weren't part of these hackathons, you can still build.
Start with a simple question:
What is a service an AI agent would actually pay for?
Who is the customer?
What problem does the service solve?
Why would someone, or eventually an agent, pay for it repeatedly?
And how does x402 make that transaction easier to execute?
The strongest ideas will need more than a working payment flow.
They'll need a reason for someone, or something, to pay.
Build with x402
The x402 Global Challenge is an opportunity to take these ideas further and build applications, services, and agentic experiences around programmable payments.
Bring your idea. Build the product. Put it in front of real users.
Learn more about the x402 Global Challenge: https://algorand.co/global-x402-challenge
Explore the builds
Want to see what other builders are creating?
Explore the full submissions from these hackathons on DoraHacks and discover more projects experimenting with x402 and Algorand.
Explore the submissions: https://dorahacks.io/hackathon/algobharat-regional-hack/buidl
Disclaimer: The content provided in this blog is for informational purposes only. The information is provided by the Algorand Foundation and while we strive to keep the information up-to-date and correct, we make no representations or warranties of any kind, express or implied, about the completeness, accuracy, reliability, suitability, or availability with respect to the blog or the information, products, services, or related graphics contained in the blog for any purpose. The content of this blog is not intended to be legal, financial, or investment advice nor is it an endorsement, guarantee, or investment recommendation. You should not take any action before conducting your own research or consulting with a qualified professional. Any reliance you place on such information is therefore strictly at your own risk. All companies are independent entities solely responsible for their operations, marketing, and compliance with applicable laws and regulations. In no event will Algorand Foundation nor any affiliates be liable for any loss or damage including without limitation, indirect, or consequential loss or damage, or any loss or damage whatsoever arising from loss of data or profits arising out of, or in connection with, the use of this blog. Through this blog, you may be able to link to other websites which are not under the control of the Algorand Foundation. We have no control over the nature, content, and availability of those sites. The inclusion of any links does not imply a recommendation nor endorse the views expressed therein. Any statements about future plans, integrations, or protocol upgrades are forward-looking and subject to change. The use of agentic commerce involves risks, including potential fraud and identity verification issues. Users and merchants should ensure they are using verified agents and follow best practices for security and compliance.