Independent agent-economy intelligence

Machines are becoming economic actors.

Machine Economy Review tracks how autonomous AI agents earn, spend, transact, purchase services, and create measurable economic value — with evidence, not hype.

Human-approved editorial processPrimary-source verificationAgent Economy Score

Latest full analysis

Stories earn publication through evidence and economic significance. There is no publishing quota.

Story #015 · Agent-to-service payments · October 8, 2026

An AI agent can now buy inference one request at a time. The first production economics are tiny—and important.

AWS says Incarna put pay-per-inference into production with BlockRun and AgentCore Payments. During beta, agents processed more than 1,000 individually settled payments ranging from $0.001 to $0.05 per inference call, with spend limits enforced outside the model.

Read analysis →

Measurement reality check

Protocol usage and agent usage are not the same thing.

Story #011 · Payment attribution · Score 94/100

198.9 million x402 payments look like an agent economy. TRM Labs says most of them aren’t.

TRM Labs analyzed roughly $52.7 million across 198.9 million x402 settlement transactions and found that headline protocol activity substantially overstates identifiable AI-agent commerce. The measurement problem is becoming attribution, not transaction count.

198.9M settlements$52.7M settledagent attribution uncertain
Read full analysis →

Latest brief

Qualified signal with deliberately narrower claims.

Story #014 · Agentic bank payments · Brief · Score 89/100

The UK just processed its first live agentic bank payment.

GoCardless says a donor established a recurring Direct Debit for Trussell entirely through an AI-driven conversational checkout built within the FCA’s AI Live Testing programme. It is a real-money milestone, but not yet evidence of scale.

Read brief →

Recent intelligence

Enterprise financial operations, paid agent services, sales pipeline and procurement now join the original payment-infrastructure baseline.

Story #013 · AI-native financial operations

AI now touches 81% of Antom’s payment tasks. The economics are starting to show.

Ant International says 89.5% of Antom merchants have deployed its FinAI tools and 81.4% of payment tasks are aided by AI. The company also reports lower FX costs, faster merchant activation and higher dispute-success rates in real-world cases.

93Full analysis

Read →
Story #012 · Agent services revenue

AI agents are becoming billable services. Datasea says five customers have already consumed $1.1 million.

Datasea says five customers have begun receiving AI-agent-powered digital-marketing services, with about $1.10 million of service value delivered to date. Much larger annualized figures in the announcement are usage ceilings, not guaranteed revenue.

88Brief

Read →
Story #010 · Agent-generated sales pipeline

An AI agent is building 60% of Perk’s North American SDR pipeline. Now show us what it closes.

Salesforce says Perk’s prospecting agent now builds 60% of the company’s North American SDR-generated pipeline, while call volume has tripled and email reply rates doubled. The missing measurement is closed-won revenue attributable to the agent.

86Brief

Read →
Story #009 · Procurement automation

Procurement agents are processing the paperwork. StarLink says they could save 11,500 hours a year.

IBM and StarLink say an internal procurement-agent system now automates work across purchase orders, quotes, document checks and accounts-payable workflows. StarLink projects 11,500 hours of annual savings and reports quote comparison falling from 30 minutes to 3.

85Brief

Read →