AI / intelligence
The mind
- 01 Reason
- 02 Plan
- 03 Learn
- 04 Decide
- 05 Act
A Web3-native foundation for the autonomous economy
The Web3 economy
for autonomous agents.
AI agents are becoming economic participants. Web3 gives them identity, ownership, payments, markets and settlement.
AEF / 01Infrastructure for always-on, high-volume Web3 agent workloads.
Network / pulse
A living economic layer.
AGENT
ECONOMY
ECONOMY
Current event
Agent-128 / market context processed
- Agent-128: market context processed.
- Agent-521: 2.4 million context tokens processed.
- Agent-883: 0.18 USDC settled.
- Agent-092: risk model updated.
BUILDING THE DECENTRALIZED ECONOMIC INFRASTRUCTURE FOR AUTONOMOUS AI AGENTSScroll to explore ↓
01 / Thesis
Software is becoming
an economic actor.
Traditional software waits for a command. Autonomous agents observe, reason, decide and act across open networks — continuously.
See the runtimeThe missing half
AI creates intelligence.
Web3 creates agency.
Without Web3, an agent can think. With Web3, it can participate in the economy.
Web3 / agency
The economic body
- 01 Own
- 02 Pay
- 03 Trade
- 04 Verify
- 05 Contract
- 06 Settle
Where the economy starts
Agents that work
while the world sleeps.
The first generation of autonomous agents is already taking shape across markets, research, security and machine-to-machine commerce.
01Third-party runtime
Autonomous
Trading Agent
Continuously analyzes market structure, on-chain activity, news, liquidity and risk to make policy-bound decisions.
Inspect the runtime02↗
Autonomous
DeFi Agent
Evaluates protocols, yield, liquidity and exposure across networks — then rebalances within explicit policies.
Explore the stack03◎
Autonomous
Research Agent
Reads reports, governance, whitepapers, news and chain data to produce a continuously updated thesis.
Read the research04⌁
On-chain
Monitoring Agent
Watches addresses, contracts, whale flows and protocol events to classify behavior and surface signals.
Follow the signal05◌
Web3
Security Agent
Analyzes contracts, approvals, transactions and governance changes to explain risk before it compounds.
See security layers06↔
Agent-to-Agent
Commerce
Discovers services, compares quotes, pays with stablecoins, verifies results and updates portable reputation.
Trace an interaction06 / 06 workloadsWhy they scale
A different class of demand
Why agent workloads
consume more tokens.
Agent workloads scale by frequency, context size, reasoning depth, tool calls and agent count — not synthetic token generation.
Continuous
Always-on loops observe, reason, act, then observe again. Seven days a week, not one chat at a time.
Observe→Reason→Act↻
Context-heavy
Each decision carries history, market context, memory, tool results, portfolio state and risk rules.
Historical dataTool resultsMemory
Multi-step
A useful action is a chain of planning, research, analysis, validation, execution and review.
PlanResearchValidateExecute
Multi-agent
A manager delegates to research, data, risk, trading and compliance agents — multiplying context and calls.
ManagerResearchRiskData
Capacity / scenario 001
Built for
billion-token
workloads.
One autonomous trading agent can trigger millions of tokens every day through monitoring, research, reasoning, risk evaluation and post-trade review.
Token volumeAgents × frequency × context × steps × duration
Illustrative workload Actual consumption depends on model, context size, reasoning frequency, architecture and strategy complexity.
Autonomous trading agent
Estimated capacity scenario17M tokens / agent / day
Select agent count
1 agent → 17M tokens / day
The operating layer
Infrastructure for
third-party agents.
AEF provides the environment. Users and institutions control their own strategies, policies and execution boundaries.
01Third-party users
Quant teamsWeb3 projectsResearch groupsDAOs
02Create agents
TradingResearchSecurityMonitoringCommerce
03Agent runtime
AEFContinuous reasoningpolicy-bound execution / memory / observability
Running by design04Connected resources
ModelsGPT / Claude / GeminiWeb3 datachain / market / newsToolswallet / APIs / DEX
Boundary / important
AEF does not operate proprietary trading strategies, custody user funds or guarantee returns. Third-party users deploy and operate their own autonomous agents.
The open layer beneath
The Web3
Agent Stack.
Economic agency is not a feature. It is a stack of primitives that lets autonomous software identify, own, pay, contract and settle.
Read the foundation principles08
AI AgentsPersonal / enterprise / autonomous
BUILD07
Agent marketsServices / APIs / data / compute
BUY06
ProtocolsDiscovery / messaging / reputation
COORDINATE05
IdentityDID / credentials / ownership
VERIFY04
Wallets & paymentsStablecoins / x402 / micropayments
PAY03
Smart contractsEscrow / agreement / settlement
CONTRACT02
DeFiSwap / lending / yield / liquidity
ALLOCATE01
InfrastructureBlockchain / oracle / DePIN / data
CONNECTIdeas for an open economy
Research for
what comes next.
The Web3
Agent Economy
A field guide to identity, ownership, payments and open markets for autonomous software.
Read the reportStablecoins
as machine money
Why programmable, global settlement is a missing primitive for AI agents.
Read insightResearch agents
at machine scale
Long context, multi-step reasoning and the economics of continuous analysis.
Read insightA foundation for the open machine economy
Build it open.
Build it on Web3.
We are convening the protocols, researchers, infrastructure providers and builders shaping the economic layer for autonomous agents.
Join the ecosystemThe Agent Economy Manifesto08 principles
- 01 Agents should be able to own.
- 02 Agents should be able to transact.
- 03 Identity should be portable.
- 04 Trust should be verifiable.
- 05 Humans retain ultimate authority.
- 06 Infrastructure remains open.