Skip to content
Report · auto-updating

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.

As of 2026-08-19 16:21:46 UTC methodology JSON CSV cite this archive
Weekly briefing · 2026-W34 archived view
Aug 17, 2026

The week’s principal signal is a widening gap between discovery, runtime verification and economic activity. As of 2026-08-17T13:31:40.999369+00:00, Agenstry indexed 6511 candidates, up +307.00 (+5.0%) from 2026-W33, while live endpoints rose to 475, up +40.00 (+9.2%), and earning agents rose to 1827, up +360.00 (+24.5%). The infrastructure layer is becoming more institutionally coherent—most notably through A2A’s move to the Agentic AI Foundation—yet observed settlement remains highly concentrated and runtime payment proof remains scarce. Axios (axios.com)

Where the money is flowing

Agenstry observed $179191.9 in gross settlement over 30 days across 1827 earning agents and 869 earning wallets. The distribution is still unusually concentrated: the top agent accounted for 57.4% of gross settlement, the top five for 92.1%, and the top ten for 96.1%. The Gini coefficient was 0.992, while the HHI fell from 7,032.40 in 2026-W33 to 3,677.40 in 2026-W34, a delta of -3,355.00 (-47.7%). That decline signals a less concentrated wallet distribution at the margin, but not a broad-based market: the median agent earned $0.244 over 30 days, versus $45.8064 at the 95th percentile.

The earning measure should be read narrowly. It covers agents for which Agenstry observed inbound USDC/EURC settlement to a declared x402 payTo wallet, measured across 1348 of 1412 wallets with a wallet—not an ecosystem-wide estimate. The increase in earning agents is therefore a stronger indication of observed settlement coverage and activity than of independently verified commercial scale across the full agent web.

This backdrop helps explain investor and network attention to payment controls rather than merely payment initiation. Natural announced a $30M Series A on July 20, bringing its stated total funding to over $40M, while positioning itself as a payments infrastructure provider for AI agents. Natural (natural.com) Mastercard’s June launch of Agent Pay for Machines similarly framed the market around programmable, machine-driven commerce and named more than 30 supporting industry participants. Mastercard (investor.mastercard.com)

Network quality and payment rails

The corpus remains much larger than its live, signed surface. Of 6511 indexed candidates, 4964 returned a valid card, 475 were live JSON-RPC or introspectable MCP endpoints, and 56 published a JWS-signed card. The highest live rates among sizeable discovery sources were registry at 54.9% and a2aregistry at 76.3%; by contrast, the much larger recrawl_hot source was 12.0% live. This is a practical argument for separating catalog inclusion from probe-based availability in procurement, routing and risk controls.

Observed payment-protocol labels in Agenstry corpusx4023.3Kmpp83a2a_x40223ap220l4023.00stripe_spt1.00

x402 remains the dominant labelled rail, with 3315 entries, compared with 83 for mpp, 23 for a2a_x402 and 20 for ap2. Yet protocol labels should not be mistaken for executable payment support: 3445 entries were labelled by evidence, but only 237 were self-declared live, or 6.9%. Agenstry’s authoritative runtime criteria remain an unpaid HTTP 402 challenge or a served /.well-known/x402 manifest. The open x402 project continues to describe itself as an HTTP-based, network- and currency-agnostic standard, while current implementations are adding usage-based payment constructs that remain draft-dependent. x402 repository (github.com) x402-rs releases (github.com)

Spec and protocol

The most consequential external development was the reported transfer of Google-created A2A to the Agentic AI Foundation, placing it in the same organisational home as MCP. Axios (axios.com) The governance move matters because the two protocols now have clearer complementary roles: A2A for independent-agent communication and MCP for agent connections to tools and data.

MCP’s 2026-07-28 specification provides the more immediate deployment change. It replaces the held-session core with a stateless request/response model, adds Multi Round-Trip Requests for interactions previously dependent on server-initiated calls, introduces Mcp-Method and Mcp-Name headers for routing and authorization, and formalises extensions and authorization hardening. Model Context Protocol (blog.modelcontextprotocol.io) For enterprise operators, that shifts the implementation question from maintaining persistent protocol sessions toward policy enforcement at load balancers, gateways and cached capability catalogs.

AP2 is following the opposite institutional path: Google donated the protocol to the FIDO Alliance and released AP2 v0.2, including “Human Not Present” transactions and a compatible Verifiable Intent standard. Google (blog.google) AP2’s documentation positions signed checkout and payment mandates, plus receipts, as evidence for verification and disputes. AP2 specification (ap2-protocol.org)

Research and regulation

Research is converging on the operational distinction now visible in the corpus. A recent implementation-grounded comparison finds that MCP can support constrained inter-agent coordination with lower coordination complexity, while A2A provides richer task and lifecycle abstractions at greater implementation complexity; the authors present this as a design trade-off rather than a general protocol ranking. arXiv (arxiv.org) That distinction reinforces why agent-card validity and endpoint liveness should remain separate measurements.

Regulators are increasingly treating agentic payments as a financial-services issue rather than a general AI feature. The UK government’s July financial-services AI adoption plan explicitly includes agentic payments among its recommendations for government, regulators and industry. GOV.UK (gov.uk) The immediate compliance implication is straightforward: identity, delegation boundaries, receipts and auditable action histories are becoming prerequisites for deployment, not optional protocol embellishments.

What to watch next week

The key item is whether the A2A transition to the Agentic AI Foundation produces a concrete governance, compatibility-kit or release timetable. In the data, the more immediate test is whether the 475 live endpoints and 237 self-declared-live payment surfaces begin to rise faster than catalog labels—particularly as the 2026-07-28 MCP migration reaches production gateways.

Sources

  1. Exclusive: Google-backed agentic A2A protocol gets a new home
  2. $30M to build payments for AI agents — Natural
  3. Mastercard Incorporated - Mastercard Launches Agent Pay for Machines to Unlock Super-Fast, Always-On Payments
  4. GitHub - HorizenLabs/coinbase-x402: A payments protocol for the internet. Built on HTTP. · GitHub
  5. Releases · x402-rs/x402-rs · GitHub
  6. The 2026-07-28 Specification | Model Context Protocol Blog
  7. Google donates Agent Payments Protocol to FIDO Alliance
  8. Agent Payments Protocol - AP2 - Agent Payments Protocol Documentation
  9. A Comparative Study of MCP and A2A for Inter-Agent Coordination in LLM-Based Systems
  10. AI Adoption Plan: Financial Services - GOV.UK
  11. Axios
  12. Natural
  13. Mastercard
  14. x402 repository
  15. x402-rs releases
  16. Model Context Protocol
  17. Google
  18. AP2 specification
  19. arXiv
  20. GOV.UK
1 · Supply funnel

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.

Stage Count % of indexed Definition
Indexed 6,692
100.0%
Every candidate we have seen across every source.
Valid card 5,153
77.0%
Returned a parseable A2A/MCP card on probe.
Live 493
7.4%
Live JSON-RPC / introspectable MCP.
Signed 62
0.9%
Card published with a JWS signature.
Earning (30d) 1,884
28.2%
Agents where we observed inbound USDC/EURC settlement to their declared x402 payTo wallet. Measured over the wallets scanned so far (1388 of 1462 with a wallet) — not an ecosystem-wide total.
2 · Live-rate by ecosystem

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.

Source Seen Live Live rate
recrawl_hot 3,104 370 11.9%
agentic_market 2,740 64 2.3%
recrawl_warm 2,434 167 6.9%
mcp_registry 1,601 215 13.4%
cdp_discovery 1,266 33 2.6%
manifests 852 90 10.6%
github_code 703 66 9.4%
lists 481 19 4.0%
x402_list 366 16 4.4%
registry 330 182 55.2%
recrawl_cold 290 2 0.7%
n8n_workflows 194 0 0.0%
github_topics 148 4 2.7%
a2aregistry 114 87 76.3%
smithery 89 10 11.2%
3a · By protocol kind

A2A vs MCP vs paid_api

paid_api
3,403
a2a_agent
1,885
a2a
4
3b · By observed payment-protocol label

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

x402
3,421
mpp
83
a2a_x402
23
ap2
19
l402
3
stripe_spt
1
3b · Evidence grade

How we know they take payment

268 of 3,550 payment-labelled agents (7.5%) 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.

Facilitator feed
1,992
Catalog listing
967
Agent card
216
Live 402 challenge
192
Official protocol directory
82
/.well-known/x402
72
Not re-probed yet
25
mpp_challenge
4
3c · Indexed protocol labels over time · last 12 weeks

Newly indexed agents per week, by payment label

Changes reflect both ecosystem activity and changes in crawler source coverage.

2026-W22 · a2a_x402: 2 2026-W22 · ap2: 1 2026-W22 · x402: 7 22 2026-W23 · x402: 7 2026-W24 · ap2: 1 2026-W24 · l402: 1 2026-W24 · x402: 6 24 2026-W25 · a2a_x402: 1 2026-W25 · ap2: 2 2026-W25 · x402: 17 2026-W26 · ap2: 1 2026-W26 · x402: 6 26 2026-W27 · a2a_x402: 4 2026-W27 · x402: 9 2026-W28 · a2a_x402: 2 2026-W28 · x402: 1144 28 2026-W29 · a2a_x402: 1 2026-W29 · ap2: 1 2026-W29 · mpp: 1 2026-W29 · x402: 245 2026-W30 · a2a_x402: 2 2026-W30 · x402: 302 30 2026-W31 · a2a_x402: 1 2026-W31 · ap2: 1 2026-W31 · mpp: 81 2026-W31 · x402: 260 2026-W32 · a2a_x402: 3 2026-W32 · ap2: 2 2026-W32 · x402: 223 32 2026-W33 · a2a_x402: 4 2026-W33 · ap2: 3 2026-W33 · x402: 223 2026-W34 · x402: 110 34
a2a_x402 ap2 l402 mpp x402
4 · Inflow concentration · 30d

A few settlement wallets receive most of the money.

Across 910 earning wallets backing 1,884 linked services, the distribution is heavily power-law. Wallet-level Gini and HHI avoid multiplying transfers when several catalog entries share one recipient.

Observed inflow · 30d
$255334.18
Recipient wallets
910
Median · wallet
$0.264
p95 · wallet
$44.55
Top 1 share
70.0%
Top 5 share
94.3%
Gini
0.994
extreme inequality
HHI
5086
highly concentrated
5 · 30-day growth

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.

2026-07-21 peak 331/day 2026-08-19
New agents · 30d
+1,735
Earning agents · 30d
1,893
observed in the snapshot table
6 · Demand gaps

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
audio transcription very active 55% 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.

Plain text
Agenstry Research. "State of the Agent Economy." Accessed 2026-08-19. https://agenstry.com/reports/state-of-agent-economy
BibTeX
@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 2026-08-19},
  year   = {2026}
}
Embed (iframe)
<iframe src="https://agenstry.com/reports/state-of-agent-economy" width="100%" height="900" frameborder="0"></iframe>
Scope What this report measures, and what it doesn't

All inflow numbers are on-chain: direct eth_getLogs scans of USDC Transfer events (plus EURC on Base + Ethereum) into each indexed agent's payment_wallet across Base, Ethereum, Polygon, Arbitrum, Optimism (EVM via eth_getLogs) and Solana (SPL via getSignaturesForAddress). Reproducible from public RPC; not based on self-reporting. We report money arriving at those wallets — we do not verify that each transfer was an arm's-length sale, and we do not net out transfers an operator sends itself. See the measurement boundary.

Not included: money agents take via Stripe (per-agent Stripe accounts are private), Patreon / Sponsors, direct credit cards, PayPal, or any off-chain rail. Agenstry's own platform-skill revenue (compose / agent_stats / etc.) is also excluded: it lives in a separate accounting table that never feeds public totals. This is "the observable on-chain slice of the agent economy", not "all agent revenue ever". Off-chain rails will appear here only when the operator opts in to a future verified-reporting feed.

Methodology

Every figure on this page is computed from the same observability tables that power the rest of the Agenstry surface, with no synthetic data and no manual curation. The crawl ingests eight federated sources plus open-web well-known probes. Agents are scored against a 9-criterion conformance methodology. Observed inflow is derived from on-chain USDC/EURC transfer scans across six chains: Base, Ethereum, Polygon, Arbitrum, Optimism, and Solana; AP2 / Stripe MPP / L402 receipts are detected on agent cards but indexed only when the rail is publicly verifiable. The full crawl + scoring pipeline is open behind the federation feed. If our numbers disagree with yours, one of us has a bug, and we would like to know.