State of the agent economy
Live measurement of the agent web: supply funnel, live-rate by ecosystem, protocol adoption, inflow concentration, and 30-day growth. Data refreshes every five minutes from the same observability spine that powers every public Agenstry page.
The agent economy entered August with a sharper split between discovery, runtime availability, and monetisation. Agenstry’s indexed corpus reached 5,855, up 19.1% from 4,918 in 2026-W30, while live agents rose 20.2% to 386; the more consequential change was in observed settlement, where earning agents increased from 21 to 1,060 as wallet coverage expanded. At the same time, revenue concentration became materially more severe: the Gini coefficient reached 0.964, with the top agent accounting for 72.1% of gross 30-day volume.
Where the money is flowing
The corpus now records $142,593.8 in gross 30-day settlement across 1,060 earning agents, but the distribution is closer to a winner-take-most market than a broad base of independent suppliers. The median agent received $1.53, compared with $124.726 at the 95th percentile; the top five agents captured 83.1% of volume and the top ten 83.9%. HHI reached 5,241.0, up from 1,495.80 in 2026-W30.
The week-over-week jump in earning agents should not be read as equivalent to a 49-fold expansion in underlying demand. Agenstry states that settlement was observed over 1,314 of 2,218 wallets with a wallet, and the metric is therefore measured over scanned wallets rather than the ecosystem as a whole. The data nevertheless establish a useful distinction for investors and regulators: monetisation is now observable at scale, but it is highly concentrated and still sensitive to measurement coverage.
The payment layer remains dominated by x402, with 2,879 records, versus 82 for MPP, 17 for AP2, and 2 for L402. The newest protocol cohort, 2026-W32, contains 17 records, all x402, while the preceding 2026-W31 cohort included 81 MPP records alongside 257 x402 records. That suggests experimentation with alternative payment rails, but not yet a durable shift in the installed base.
Spec & protocol
The most significant external protocol event was the July 28, 2026 MCP specification release. The release introduces a stateless operating model over commodity HTTP, formalises an extensions track, and adds a Tasks lifecycle in which clients can create, inspect, update, and cancel long-running work through task handles. The official MCP announcement describes the release as a stability point for implementers building against the 2026-07-28 specification. MCP specification release
A2A also moved closer to a more mature governance and implementation phase. The official A2A community hub flagged an upgrade on July 26, while Oracle documented a July 7 capability allowing external A2A-compatible clients to discover and call Select AI agent teams through standard HTTPS interfaces. A2A community updates Oracle A2A release note
For Agenstry, the practical implication is that conformance testing will increasingly need to distinguish between a static card, a reachable endpoint, and a service that correctly implements asynchronous or stateful protocol semantics. The current funnel makes that gap visible: 4,307 entries have a valid card, but only 386 are live, and just 50 publish a JWS signature.
Notable launches and capital
Enterprise adoption continued to move through infrastructure and control planes rather than consumer-facing autonomous workers. Abrigo announced its Agentic Platform Experience for financial institutions, targeting lending workflows and governance requirements, with a Q3 2026 launch plan. Abrigo Agentic Platform Experience
Security and observability attracted a notable financing event: Neo emerged from stealth with $100 million backed by Andreessen Horowitz and Bessemer Venture Partners, positioning itself around inventory, attribution, posture intelligence, and policy enforcement for agentic software. Neo funding announcement
The pattern is consistent with Agenstry’s measurements. The ecosystem is adding inventory rapidly—2,651 new agents in the last 30 days, compared with 1,834 in the prior comparison period—but the scarce asset is not listing volume. It is verified runtime access, payment identity, attribution, and evidence that an agent can operate predictably inside enterprise controls.
Research
Recent research is shifting from abstract coordination toward institutional infrastructure. A July paper, Underwriting the Agent Economy: The Blueprint for an AI Insurance Stack, frames insurance as a missing layer for systems that transact and act with limited human intervention. Underwriting the Agent Economy
A separate July paper on regulating autonomous and agentic AI compares regulatory approaches across content platforms, data protection, financial services, and the EU AI Act. Regulating autonomous and agentic AI The common thread is operational traceability: who delegated an action, which agent executed it, what tools and counterparties were involved, and whether a human could intervene.
Regulation
The EU moved from framework-setting to operational enforcement. The European Commission published guidance on transparency obligations that began applying on August 2, 2026, and its AI Act Service Desk states that agents intended to interact with natural persons or generate content fall within the relevant transparency rules. European Commission transparency guidance EU AI Act Service Desk on agents
The policy burden is therefore moving toward evidence. Agenstry’s distinction between self-declared live status—221 entries, or 7.4% of labelled records—and runtime challenges is increasingly aligned with what regulated deployments will require: machine-readable identity, auditable action histories, and independently testable claims about capability.
What to watch next week
The key signal will be whether the MCP 2026-07-28 release produces a measurable increase in live, task-capable endpoints rather than another wave of static declarations. Agenstry should also track whether the recent MPP cohort persists after 2026-W31 or reverts toward the x402-heavy pattern visible in 2026-W32.
Most of the agent economy is indexed, not yet operational.
Counting agents conflates supply with operation. Below is the corpus broken into five stages: what fraction of indexed agents pass each gate. The decay curve is the honest read on how mature the agent web is at any given moment.
Which source registries actually ship live agents.
Each row shows one source registry (Anthropic's MCP directory, Smithery, Glama, Postman, GitHub well-knowns, Wayback CDX scans) alongside how many of the agents it lists actually respond live on probe. The gap between seen-count and live-count is the source's effective freshness.
A2A vs MCP vs paid_api
Index composition, not global market share. x402 has multiple dedicated catalogs and a live HTTP probe; AP2, L402 and Stripe ACP are predominantly visible only when cards declare them.
x402 / AP2 / Stripe MPP / L402
How we know they take payment
3,615 of 7,201
payment-labelled agents (50.2%) proved it themselves —
their own endpoint answered an unpaid request with an HTTP 402 challenge, or the host
serves a /.well-known/x402 manifest. The rest is declared in a static
agent card or described by a third-party catalog.
Newly indexed agents per week, by payment label
Changes reflect both ecosystem activity and changes in crawler source coverage.
A few settlement wallets receive most of the money.
Across 2,159 earning wallets backing 4,801 linked services, the distribution is heavily power-law. Wallet-level Gini and HHI avoid multiplying transfers when several catalog entries share one recipient.
New agents arriving daily, by first-seen.
Daily count of agents we discovered for the first time. Sourced from every registry + open-web crawl we run.
What the market asked for and didn't get.
Every section above measures the supply we indexed. This one measures demand we couldn't serve: clustered search intents — from browsers, AI assistants and agents calling our A2A and MCP endpoints — where the index returned nothing or almost nothing. Aggregate only, and shown only for intents searched by enough distinct sources to be a market signal rather than one person's session.
| Intent | Demand | Unmet | Trend |
|---|---|---|---|
| aidata | active | 42% | new |
| bitcoin+payment | very active | 38% | rising |
Full list + method: what the agent economy is missing. Building one of these? Add pay-per-call payments to your API →
Cite this report
Public domain numbers (CC BY 4.0). Pick the format that fits. The page auto-updates so the URL is the canonical pointer; the date stamps the snapshot you cited.
Agenstry Research. "State of the Agent Economy." Accessed 2026-11-17. https://agenstry.com/reports/state-of-agent-economy
@misc{agenstry_state_of_agent_economy,
title = {State of the Agent Economy},
author = {Agenstry Research},
url = {https://agenstry.com/reports/state-of-agent-economy},
note = {Accessed 2026-11-17},
year = {2026}
}
<iframe src="https://agenstry.com/reports/state-of-agent-economy" width="100%" height="900" frameborder="0"></iframe>
Inflow figures count public stablecoin transfers into wallets declared by indexed agents. They are reproducible from public chains, but a transfer may be self-funded and is not proof of a sale. See the measurement method.
Private and off-chain payments, and Agenstry's own paid-skill receipts, are excluded. Coverage varies by chain, asset and date.
Methodology
This report uses dated observations from the public index. Check source coverage, scoring rules and the public feed before comparing it with another dataset.
Weekly State of the Agent Economy
One analyst briefing a week, Mondays: earning wallets, observed inflow, new agents and the news behind them, with sources.
That email address doesn't look right. Check it and try again.