A guide for building agentic projects on ARC Name Service (ANS) for hackathons.
Based on hands-on integration of ANS into Arc Global Payouts.
Author: GoGo — @0xGoGochain Date: April 2026
Yes, very much so.
AI agents need to send, receive, and reference money autonomously. Wallet addresses (0x858f...5f4E) are hostile to agents — they're impossible to reason about, easy to hallucinate, and can't be validated semantically.
.arc names solve this: an agent can understand alice.arc the same way it understands alice@email.com.
ANS sits at the intersection of three hot hackathon themes:
- AI agents — autonomous software that acts on behalf of users
- Stablecoins (USDC) — real-world value transfer
- Onchain identity — human-readable, verifiable naming
This combination makes ANS a strong hackathon entry with clear narrative appeal.
| Problem Without ANS | Solution With ANS |
|---|---|
Agents pass around opaque 0x... addresses |
Agents reference alice.arc — semantic and verifiable |
| Transaction logs are unreadable | Logs show paid supplier.arc 500 USDC — auditable |
| Agent config requires hardcoded addresses | Agents discover each other by .arc name |
| LLMs can hallucinate hex addresses | .arc names are short, validatable strings |
| No agent identity standard | .arc names become the agent identity layer |
The current ANS is human-facing (type a name, see it resolve). For agents, the paradigm shifts:
Every AI agent gets a .arc name as its onchain identity:
payroll-agent.arc— a company's salary disbursement bottreasury.acme.arc— an organization's treasury agentrebalancer.arc— a DeFi portfolio agent
Agents register their own names, publish their capabilities onchain, and other agents discover them by name — not by passing around raw addresses.
Instead of hardcoding addresses in agent configs:
// Without ANS — brittle, opaque
agent.send("0x858f3232E7d6702F20c4D3FEAB987A405D225f4E", "100 USDC")
// With ANS — semantic, auditable
agent.send("supplier-agent.arc", "100 USDC")
This makes agent transaction logs human-readable and auditable — critical for trust.
The strongest hackathon demo: a user speaks to an agent, the agent resolves .arc names and executes payments autonomously:
User: "Pay the design team 500 USDC each"
Agent:
→ Looks up "design-team.arc" → finds a group record with 3 members
→ Resolves alice.arc, bob.arc, charlie.arc
→ Executes 3 USDC transfers on Arc
→ Reports back: "Paid 1500 USDC to 3 members of design-team.arc"
An SDK where AI agents register .arc identities and pay each other for services.
How it works:
- Agent registers
my-agent.arcwith a metadata record describing its capabilities (e.g., "image generation", "data analysis") - Other agents discover services by querying ANS records
- Payments happen automatically via
.arcnames
Demo flow:
- Agent A needs an image generated
- Agent A queries ANS for agents with "image-generation" capability
- Finds
image-gen.arc, which quotes 2 USDC per image - Agent A pays
image-gen.arc2 USDC image-gen.arcdelivers the result- Transaction recorded:
agent-a.arc → image-gen.arc: 2 USDC (image-generation)
Tech stack: Next.js + wagmi + ANS SDK + Groq/OpenAI for agent reasoning
A multi-agent system that manages company finances using .arc names as the identity layer.
Agents:
cfo-agent.arc— approves payments above a thresholdpayroll-agent.arc— executes scheduled salaries toemployee1.arc,employee2.arc, etc.auditor-agent.arc— monitors all transactions, flags anomalies
Demo flow:
- User says: "Run payroll for April"
payroll-agent.arcfetches employee list from ANS group record- Resolves 10 employee
.arcnames to addresses - Requests approval from
cfo-agent.arc(amount exceeds threshold) cfo-agent.arcapproves after checking treasury balance- Batch payment executes on Arc
auditor-agent.arcgenerates human-readable report using.arcnames
Tech stack: Next.js + Circle AppKit + ANS + multi-agent orchestration (LangChain/CrewAI)
Extend ANS with an agent-specific metadata and reputation layer.
How it works:
- Agents register with capabilities, pricing, uptime stats stored as ANS metadata
- Other agents query: "Find me an agent that can translate documents, under 1 USDC per page"
- ANS becomes a decentralized agent marketplace
- Reputation scores stored onchain, tied to
.arcnames - Ratings accumulate from successful agent-to-agent transactions
Demo flow:
coordinator.arcreceives task: "Translate this PDF to Spanish"- Queries ANS registry for agents tagged
translation+spanish - Finds 3 candidates, ranks by reputation score
- Selects
translator-pro.arc(highest score, 0.5 USDC/page) - Pays and receives translated document
- Submits 5-star rating → onchain reputation update
Tech stack: Solidity (metadata extension contract) + Next.js + ANS + AI orchestration
A chat-based wallet where everything is .arc-native — no addresses ever shown to the user.
Supported commands:
- "Send 50 USDC to david.arc"
- "How much did I pay supplier.arc this month?"
- "Set up weekly 100 USDC to landlord.arc"
- "Who is 0x858f...?" → "That's david.arc"
- "Register my-business.arc for my wallet"
- "Show all payments to *.arc names this week"
Demo flow: Full conversation loop: ask → resolve → confirm → pay → receipt — all using names, never addresses.
Tech stack: Next.js + Groq (LLM) + ANS SDK + Circle AppKit
Note: The Arc Global Payouts AI assistant page is already halfway to this demo. Adding full ANS-native conversation would make it a complete hackathon entry.
| Extension | What It Does | Why Agents Need It |
|---|---|---|
| Metadata records | Store JSON capabilities, pricing, API endpoints alongside .arc names |
Agent discovery — "what can translator.arc do?" |
| Group names | design-team.arc resolves to multiple addresses |
Multi-agent coordination, batch payments |
| Permissioned resolution | Only authorized agents can resolve certain names | Access control for private agent networks |
| Event subscriptions | Watch for new registrations matching a pattern | Agents auto-discover new services |
| Transaction tagging | Tag payments with .arc names onchain |
Auditable agent-to-agent payment trails |
| Text records (ENS-style) | avatar, url, description, agent-api-endpoint |
Rich agent profiles queryable onchain |
| Delegation | treasury.arc delegates spending to payroll-agent.arc |
Hierarchical agent permissions |
| Expiry alerts | Notify agents before their .arc name expires |
Prevent identity loss for autonomous agents |
Use this as a starting point for your hackathon submission:
Problem: AI agents can't use wallets like humans. They need semantic identity to transact autonomously, discover services, and produce auditable payment trails.
Solution: We built [X] — a system where AI agents register
.arcnames as their onchain identity, discover each other by capability, and pay each other in USDC — all without human intervention.How it works:
- Agents register
.arcnames with capability metadata- Agents discover each other by querying ANS
- Payments route through
.arcnames — no hardcoded addresses- All transactions are human-readable and auditable
Key insight: No copy-pasting addresses. No manual bridging. Just
agent.send("alice.arc", "50 USDC").Built on: Arc Network · ANS · Circle AppKit · USDC
Why Arc? Sub-second finality, USDC-native gas, predictable fees — ideal for high-frequency agent transactions.
Most agentic hackathons evaluate on these axes. Here's how ANS-based projects score:
| Criterion | How ANS Projects Score |
|---|---|
| Innovation | Onchain identity for agents is novel — most agent projects ignore naming |
| Technical depth | Smart contract integration + LLM orchestration + real payments |
| Practical utility | Agents paying agents is the future of autonomous commerce |
| Demo quality | Natural language → name resolution → USDC payment is a visceral demo |
| Composability | ANS names work across any Arc dApp — not siloed |
| Business potential | Agent marketplace, agentic payroll, autonomous treasury — all viable products |
The Arc Global Payouts project already includes:
- ✅ USDC payments on Arc (Send, Batch, Bridge, Swap)
- ✅ AI assistant with natural language payment commands (Groq-powered)
- ✅ ANS integration (forward resolution, reverse lookup, all address inputs)
- ✅ Transaction history and analytics
- ✅ Pay links and user profiles
To turn this into an agentic hackathon entry, add:
- Agent
.arcname registration flow - Agent capability metadata (stored onchain or in ANS text records)
- Agent-to-agent discovery and payment
- Multi-agent orchestration (e.g., approval workflows)
- Human-readable transaction reports using
.arcnames
| Question | Answer |
|---|---|
| Is ANS good for agentic hackathons? | Yes — identity + payments + AI is a strong narrative |
| What's the angle? | Agents need semantic, human-readable identity to transact autonomously |
| Strongest demo? | Natural language → .arc resolution → autonomous USDC payment |
| What's missing from ANS for agents? | Metadata records, group names, agent discovery, event subscriptions |
| Best project idea for solo builder? | Idea D — Conversational Wallet (builds on existing AI assistant) |
| Best project idea for team? | Idea B — Agentic Payroll (multi-agent, impressive demo) |
| Differentiation? | Most agent projects ignore identity — ANS makes agent payments auditable and composable |
Built on Arc Network · Powered by Circle · ANS Registry: 0xaDe3b1ae4C5831163Fe8e9727645e2416DD83AD2