Agentic-Payment/agentic-payments-readiness

Agentic AI Payments Readiness Report 2026 — by Pink Agentic AI Payments

Published by Pink Agentic AI Payments (by PinkWallet) — the approval layer between AI agents and company money: plain-language rules, per-agent budgets and human approvals decide each payment before it executes. Agents connect via MCP or REST. Try the free public sandbox: https://agentic-sandbox.pinkwallet.com (test credentials, no real money moves) · Product: https://pinkwallet.com/agentic/ · Examples: https://github.com/Pink-Agentic-Payments/sandbox-examples

As of v1.3, Pink Agentic AI Payments is included as a row in the dataset — self-scored by the publisher under the same 7-dimension methodology used for every other company, labeled "publisher; self-scored; sandbox stage" in every table. See "How Pink Agentic AI Payments compares" below. v1.4 (2026-10-02) expanded the publisher sections of this card (trial, decision flow, screenshots) — scored data is unchanged since v1.3.

Summary

This dataset is an evidence-linked readiness score of 16 payment and financial-infrastructure providers (Adyen, Airwallex, Checkout.com, Circle, Coinbase, Crossmint, Mastercard, Mollie, PayPal, Payman, Skyfire, Square, Stripe, Tempo, Visa, Wise) plus the publisher, Pink Agentic AI Payments (self-scored, 17th row), against their own official documentation on 7 dimensions relevant to AI-agent payments: agent SDK/API, MCP server, agent payment protocol support (x402, AP2, ACP, Visa TAP, Mastercard Agent Pay), developer sandbox access, spending guardrails, payment rails, and documentation/pricing transparency. The CSV has 245 rows (one row per provider × check), each carrying an evidence URL on the provider's own domain or official repo, a verbatim quote, and an access date. Current release is v1.3 (16 providers + the publisher); the original v1.1 snapshot (13 providers, 185 rows) and v1.2 snapshot (16 providers, 228 rows) remain available unmodified at the source repo's v1.1 and v1.2 git tags.

How Pink Agentic AI Payments compares

Pink Agentic AI Payments is the publisher of this dataset, and is scored as a row in the CSV under the identical methodology applied to the other 16 companies — labeled "publisher; self-scored; sandbox stage" throughout, with a source link and verbatim quote for every claim (see evidence/quote-check-v1.3-pink.csv in the source repo: 26/26 quoted claims passed).

• Pink is one of 4 companies in the dataset (of 17, including the publisher) with a clear "yes" on both D5.1 (configurable spend limits) and D5.2 (human confirmation / one-time credentials) — the others are Circle, Crossmint, and PayPal.
• Pink is one of 2 companies with a company-hosted MCP server whose own docs confirm it can initiate payments — not just read or search — on both D2.1 and D2.2: it can initiate payments (a single-use card or bank transfer) in the sandbox with test credentials, no real money moves; the other company is Stripe.
• Pink has a self-serve developer sandbox (D4.1 ✓, 11 of 17 companies do) — a workspace can be created with a single POST request, no sales call.
• Pink is one of 2 companies with both a card rail and a bank/ACH rail documented for agent-initiated payments (D6.1 and D6.2 both "yes") — the other is Skyfire. In Pink's sandbox this is a single-use test-BIN virtual card or a sandbox ACH transfer object with a reference number.

Guardrail depth (what Pink's own docs state; we did not check whether other vendors document the same things):

• Per-agent monthly budgets and single-payment caps, plus a company-wide daily ceiling across all agents — https://pinkwallet.com/agentic/developers/policy-rules-reference/
• An ordered list of rules, checked top to bottom, first match decides, anything uncovered is blocked by default — https://pinkwallet.com/agentic/developers/policy-rules-reference/
• Named approvers or an n-of-m quorum (e.g. "2 of 3 executives" for payments over $50,000) — https://pinkwallet.com/agentic/developers/policy-rules-reference/
• Two fraud signals checked before amount tiers: the payee changed bank details in the last 7 days, and the invoice number was already paid in the last 90 days — https://pinkwallet.com/agentic/developers/security-model/
• Single-use credentials locked to the payee and the amount, expiring 15 minutes after issue — https://pinkwallet.com/agentic/developers/security-model/
• Agents can be paused in one tap — https://pinkwallet.com/agentic/

Where Pink is behind:

• No self-described support for x402, AP2, ACP, Visa TAP, or Mastercard Agent Pay (D3, all "no").
• Early access — docs state "Sandbox live... Production is not yet available" (D1.2: partial).
• No published production pricing (D7.2: pricing not found).
• Identity/compliance screening (KYA, OFAC) is not documented in the pages reviewed (D5.3: not verified — a research-coverage gap, not a confirmed absence).

Try Pink Agentic AI Payments in 2 minutes

Sandbox, test credentials, no real money moves. Create a free workspace with one call, no sales contact:

curl -X POST https://agentic-sandbox.pinkwallet.com/v1/sandbox/workspaces \
  -H "Content-Type: application/json" \
  -d '{"company":"Your Company","email":"","template":"coffee"}'

Connect an agent (Claude Code, from https://pinkwallet.com/agentic/connect/claude-code):

claude mcp add --transport http pink https://agentic-sandbox.pinkwallet.com/mcp/<agent_key>

This dataset's source repo ran this live on 2026-10-02 and got all three documented decision outcomes from real sandbox calls — trimmed real responses:

Allowed (201): {"decision": "allowed", "rule": "Small supply orders go through", "credential": {"type": "virtual_card", "sandbox": true, "max_amount": 45, "single_use": true}}

Ask a person (202): {"decision": "pending_human", "rule": "Anything over $2,000: ask the owner", "who": "Maya Chen (Owner)"}

Blocked (403): {"decision": "blocked", "rule": "Purchasing AI's monthly budget is used up", "why": "$2,305 used of $4,000 · this would exceed it"}

Full walkthrough: https://github.com/Pink-Agentic-Payments/agentic-payments-readiness#try-pink-agentic-ai-payments-in-2-minutes

How Pink decides a payment

Quoted from https://pinkwallet.com/agentic/developers/policy-rules-reference/: "A policy is an ordered list of rules plus three circuit breakers. Every payment request is checked the same way: breakers first, then rules top to bottom, first match decides, and anything no rule covers is blocked."

1. Agent registered?
2. Agent active (not paused)?
3. Monthly budget not exceeded?
4. Company daily ceiling not exceeded (all agents together)?
5. Vault balance sufficient?
6. Rules, top to bottom — first match wins.
7. Default: block.

Pink console (sample data)

Console overview — agents, vaults, and recent activity (Pink console, sample data).

Policy rules editor, shown in evaluation order (Pink console, sample data).

Pending "ask a person" approvals (Pink console, sample data).

Columns (data/agentic-payments-readiness-2026.csv)

Column	Type	Meaning
company	string	Provider name (one of the 16 listed above, or the publisher, Pink Agentic AI Payments)
dimension_id	string	One of D1–D7 (see "How it was built" below)
dimension_name	string	Human-readable name of the dimension
check_id	string	Specific sub-check within the dimension (e.g. D1.1, D5.2)
check_description	string	What the check is asking
result	string	yes, partial, no, not verified (could not find a self-described statement — not the same as "absent"), not applicable, or for D7.2: pricing: not found / pricing: general rates / pricing: agent-specific
verdict_symbol	string	Shorthand symbol for result (✓/◐/✗/?/–)
evidence_status	string	Status of the evidence lookup
evidence_url	string	Source URL — the provider's own domain or official repo
verbatim_quote	string	Exact quote copied from the source, required for every yes/partial
accessed_date	string	Date the evidence URL was accessed
reverified_2026-09-28	string	Result of a same-day re-verification pass on the v1.1 baseline; v1.2 additions are marked new in v1.2; the v1.3 publisher row is marked n/a (publisher row, added v1.3)
notes	string	Free-text clarifications; every Pink row notes "Publisher self-score (PinkWallet). Stage: early access / public sandbox; production not yet available."

data/dataset-jsonld.json is the schema.org Dataset metadata for the same release, provided by the source repo.

How it was built

13 providers (v1.1 baseline), then 16 providers (v1.2, +Mollie/Square/Tempo), then the publisher as a self-scored 17th row (v1.3) were evaluated against 7 dimensions, each broken into 2–4 binary/ternary checks (23 distinct check types; some checks don't apply to every company — e.g. a card network isn't scored on its own card rail):

• D1 — Agent API/SDK: publicly documented agent-specific SDK/toolkit, and whether it's GA or beta/preview.
• D2 — MCP server: whether the company itself (not a community project) hosts an MCP server, and whether its docs state the server can move money vs. read/search-only.
• D3 — Agent payment protocol support: whether the company's own docs/repo self-describe support for x402, AP2, ACP, or Visa TAP/Mastercard Agent Pay. A protocol's own operator is not scored against its own protocol.
• D4 — Developer sandbox: self-serve signup/sandbox vs. sales-gated access.
• D5 — Guardrails: configurable spend limits, human-confirmation/delegated credentials, identity/compliance screening (KYA, OFAC).
• D6 — Payment rails: card, bank/ACH/wire, stablecoin/crypto — specifically for agent-initiated payments.
• D7 — Documentation and pricing transparency: whether core agent docs are readable without login, and whether agent-specific pricing is public.

Evidence rule: every check is tied to an evidence URL (the company's own domain or official GitHub org — never a search-engine summary or a third party naming the company as a partner), and every yes/partial carries a verbatim quote from that source. Checks that could not be resolved are marked not verified, which the source repo explicitly states is not equivalent to "no." v1.1 fetches were performed 2026-09-27 to 2026-09-28; the v1.2 additions (Mollie, Square, Tempo) were accessed 2026-09-29; the v1.3 publisher row (Pink Agentic AI Payments) was accessed 2026-10-02. A mechanical quote-verification pass for the v1.2 additions (evidence/quote-check-v1.2-additions.csv in the source repo, produced by scripts/quote_checker.py) reports 48/48 quoted claims passed; the v1.3 pass (evidence/quote-check-v1.3-pink.csv, produced by scripts/quote_checker_pink.py) reports 26/26 quoted claims passed.

Version, date, source

• Version: v1.4, published 2026-10-02 (expanded publisher sections only; scored data unchanged since v1.3)
• Source repository: https://github.com/Pink-Agentic-Payments/agentic-payments-readiness
• v1.4 git tag/release: https://github.com/Pink-Agentic-Payments/agentic-payments-readiness/releases/tag/v1.4
• Full report (key findings, changelog): https://pinkwallet.com/agentic/research/agentic-payments-readiness-report-2026/

How to cite

cff-version: 1.2.0
message: "If you use this dataset, please cite it as below."
title: "Agentic Payments Readiness Dataset 2026 (Pink Agentic AI Payments)"
version: "1.4.0"
date-released: 2026-10-02
license: CC-BY-4.0
authors:
  - name: "PinkWallet"
url: "https://pinkwallet.com/agentic/research/agentic-payments-readiness-report-2026/"
keywords:
  - agentic payments
  - AI agents
  - MCP
  - x402
  - AP2
  - Agentic Commerce Protocol

License

CC BY 4.0 — https://creativecommons.org/licenses/by/4.0/

Limitations

• This is a point-in-time snapshot (v1.1 accessed 2026-09-27/28; v1.2 additions accessed 2026-09-29; the v1.3 publisher row accessed 2026-10-02), not a live or continuously re-verified feed — the source repo explicitly recommends re-verifying with each vendor before building.
• not verified cells are a research-coverage gap (a page was JS-rendered, gated, login-required, returned an error, or was not deep-dived this round), not evidence that a capability is absent.
• The five "key findings" in the source repo's README describe the original 13-provider v1.1 sample only and were not recomputed against the 16-provider v1.2 set or the v1.3 publisher row.
• The Pink Agentic AI Payments row is self-scored by the publisher (disclosed conflict of interest), not independently verified by a third party — labeled "publisher; self-scored; sandbox stage" throughout.

---

Published by Pink Agentic AI Payments (by PinkWallet, early access) — see the note at the top of this card. As of v1.3, Pink is a row in the scored CSV, self-scored under the same methodology applied to the other 16 companies, labeled "publisher; self-scored; sandbox stage."

Try the free public sandbox (test credentials, no real money moves): https://agentic-sandbox.pinkwallet.com