Skip to content

ValenteCreativo/ARVI

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

97 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

ARVI — Agentic Regeneration Via Intelligence

Status Base Venice ERC-8004 License

Four years of research. One working loop. Zero humans required.

Environmental data is trapped in government servers. Crises are detected weeks late. The people closest to the ecosystems — farmers, park rangers, community stewards — have no tools, no incentives, and no voice in the data economy.

ARVI gives them all three.


Prove It in 15 Seconds

curl https://arvi-eight.vercel.app/api/demo | jq .

That single command runs the full chain: real weather data → Venice AI analysis → anomaly detection → alert written to Cloudflare R2 → proof artifacts returned. No mock. No simulation. Every field is verifiable.


What Happens in That Call

Step What runs Proof
1 Open-Meteo API fetches real sensor coordinates lat/lon + live weather in response
2 Venice AI (llama-3.3-70b) analyzes privately, zero retention "model_used": "venice-llama-3.3-70b", "simulated": false
3 Anomaly confirmed → email alert dispatched "email_sent": true
4 Alert written to permanent public log on Cloudflare R2 alert-log.json
5 On-chain event emitted via ARVIAgent on Base Mainnet verified tx on Basescan

This loop runs automatically every hour. No human triggers it.


The Problem It Solves

Forests burn for days before anyone responds. Wetlands dry out over months with no one noticing. Air quality spikes go undetected in neighborhoods without expensive monitoring stations.

The existing system is centralized, slow, and built around waiting for humans to act.

ARVI flips the model: citizens deploy sensors and earn passive income. Venice AI analyzes the data privately and continuously. The agent acts the moment an anomaly is detected — alerting, logging, coordinating — without waiting for permission.


Live Proof Artifacts

Artifact Link
🗂️ Live alert log (R2, permanent) alert-log.json
📋 Agent manifest (ERC-8004 / x402) arvi.skill.md
🔗 ARVIAgent on Base Mainnet Basescan
🤖 ERC-8004 on-chain identity AgentId 33311
🔁 Open mission board /api/missions
🌐 Live app arvi-eight.vercel.app

The Business Model

Who pays the rewards? Not grants. Not speculation. Three real buyers:

Who pays What they buy Price range
Cities & municipalities Hyperlocal environmental data subscriptions $300–500/month per 10 nodes
Corporations (ESG) Continuous, auditor-grade environmental evidence $2,000–10,000/month
Impact projects (Octant, Gitcoin) MRV cycles for carbon credits & impact funding Per evaluation

NGOs and government agencies add periodic intelligence reports ($300–1,500/month). Other AI agents pay per-query micropayments via x402 on Base.

The flywheel: more sensors → more coverage → more enterprise contracts → more revenue → more rewards for operators → more sensors deployed.

Sensor operators earn variable passive income based on data quality, uptime, and location scarcity. Early network nodes earn more (scarcity premium). This is infrastructure you own, not a yield farm.


The Research Behind It

ARVI is not a hackathon idea. It is the fourth iteration of a Master's thesis in Strategic Business Administration — an experiment-based research arc that started in 2023 with documentary research on DePIN for environmental impact.

Version Year What was learned
Rembu 2024 Economic incentives can drive citizen sensor adoption (DePIN validation)
SEN 2025 Hyperlocal environmental data has real market demand
Aona 2025 AI can replace human analysts for anomaly detection
ARVI 2026 Autonomous agents can close the full loop — detect, decide, act

Each version ran in a real hackathon. Each validated one hypothesis. ARVI is where they converge.


Sponsor Integrations

🟣 Venice AI — Private Inference

Sensor data is sensitive. Species locations, ecosystem vulnerabilities, proprietary monitoring zones — none of it should train a cloud model. Every ARVI analysis runs on Venice with zero retention. Every response includes "simulated": false.

🔵 Protocol Labs — No Humans Required + ERC-8004

The agent detects → decides → acts → logs, hourly, with no human trigger. It has a verifiable on-chain identity: AgentId 33311, registered at eip155:8453:0x8004A169FB4a3325136EB29fA0ceB6D2e539a432.

🟢 Octant — Data Collection and Analysis for Impact Project Evaluation

Impact projects claim outcomes that no one can independently verify. ARVI gives them continuous, sensor-backed, AI-analyzed evidence — not self-reported. Any Octant-funded project can deploy a node and get an objective impact score.

🔷 ENS — Operator Identity

Node operators register a .eth name once. Every alert, payment, and mission resolves to that name — no raw addresses. The ENS endpoint is live:

curl https://arvi-eight.vercel.app/api/ens?name=vitalik.eth

🟡 Base — Settlement Layer

Operator rewards, inference micropayments, and alert events all settle on Base. ARVIAgent contract: 0x8118069E26656862F8a0693F007d5DD7664Acb00


Stack

Frontend     Next.js 15 · TypeScript · Tailwind · Framer Motion
LLM          Venice AI (llama-3.3-70b) — private, zero retention
Blockchain   Base Mainnet (ARVIAgent) · ERC-8004 identity
Alert log    Cloudflare R2 — public, permanent, append-only
Email        Resend — medium/high/critical anomaly alerts
Data         Open-Meteo API (real-time weather + environmental)
Cron         Vercel — hourly autonomous loop
Identity     ENS — operator and node addressing

ARVI — Pantera Labs 2026 · Built for The Synthesis Hackathon

From México, with 💚.

About

Agentic Regeneration Via Intelligence -- Synthesis Hackathon 2026

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors