Making the market make sense
Plain, useful writing on agentic payments — for the developer wiring up their first agent and the newcomer who doesn’t yet know what to ask.
Policy & trustUpdated Oct 2026, 7min
Why an agent needs more than a credential
Stripe Link can now give agents safe, single-use credentials. The next question is whether an agent needs a wallet approval for every purchase or its own bounded account and policy layer.
MarketUpdated Oct 2026, 15min
The AI agent payments landscape in 2026
Which payment rails are ready for autonomous AI agents in 2026? A map of wallets, identity layers, card issuers, execution stacks, and network protocols, updated for Stripe Link's expansion into Muse, Instinct, and Grok Bot.
ProductAug 2026, 7min
Does your agent need its own credit card?
Should an AI agent share your card, get dedicated agent cards, or run from an agent account? A 2026 comparison of blast radius, attribution, and the spend that cards never see.
EngineeringUpdated Oct 2026, 10min
Which rail should your agent pay on?
A practical comparison of the four rails AI agents transact on in 2026: cards, bank transfers, stablecoins and x402, and API-native billing. Where each wins, where each fails, and why the choice should be per transaction.
EngineeringAug 2026, 7min
How to set spending limits on AI agents
A practical guide to hard spending limits for AI agents: the four limit layers, approval thresholds, rail-aware enforcement, and the calibration loop that tightens risk without killing throughput.
EngineeringAug 2026, 6min
The agent spending security checklist
A production security checklist for AI agent payments: funding isolation, intent gating, credential handling, tool scoping, hard controls, kill switches, evidence logs, and the drills to run before launch.
Policy & trustAug 2026, 7min
How agent payments get verified
What payment verification actually means for AI agents: proving which agent acted, under what authorization, with what intent, and that the transaction matched. The five layers, and what verified looks like in production.
Policy & trustUpdated Oct 2026, 8min
Who pays when an agent gets it wrong?
Chargeback rules assume a human made the purchasing decision. When an AI agent buys the wrong thing, liability gets murky fast. The 2026 landscape, the evidence that changes dispute outcomes, and the refund loop that avoids disputes entirely.
EngineeringUpdated Oct 2026, 8min
Making every agent charge explainable
Agent payment systems win by making every purchase explainable. A 4,800 dollar overspend scenario, the evidence record that survives audits, receipt capture, and the explain-this-charge workflow.
EngineeringAug 2026, 8min
How to let Claude Code spend money safely
Claude Code can register domains, buy API credits, and pay for infrastructure. Here is the pattern that makes agent purchases safe: an MCP payment tool, policies outside the prompt, scoped credentials, and human approvals.
EngineeringAug 2026, 8min
Letting OpenClaw agents spend money safely
How to give OpenClaw agents payment capabilities without lighting money on fire: what a good payment skill looks like, permission scoping, the intent-to-verify checkout flow, the control stack, and subscriptions that never expose a main card.
EngineeringUpdated Oct 2026, 8min
Letting Hermes agents spend money safely
Hermes became 2026's fastest-growing agent harness, and payments are an add-on. What Stripe skills, plugins, and soft caps cover today, why persistent memory raises the stakes, and the four-piece pattern that lets a Hermes agent spend safely.
MarketUpdated Oct 2026, 8min
How big are agent payments, really?
How big is the AI agent payments market in 2026? Measured commerce signals, forecast ranges from Grand View to McKinsey and Morgan Stanley, network rollout data, and why the estimates diverge so widely.
Policy & trustUpdated Oct 2026, 7min
Why an agent needs more than a credential
Stripe Link can now give agents safe, single-use credentials. The next question is whether an agent needs a wallet approval for every purchase or its own bounded account and policy layer.
MarketUpdated Oct 2026, 15min
The AI agent payments landscape in 2026
Which payment rails are ready for autonomous AI agents in 2026? A map of wallets, identity layers, card issuers, execution stacks, and network protocols, updated for Stripe Link's expansion into Muse, Instinct, and Grok Bot.
ProductAug 2026, 7min
Does your agent need its own credit card?
Should an AI agent share your card, get dedicated agent cards, or run from an agent account? A 2026 comparison of blast radius, attribution, and the spend that cards never see.
EngineeringUpdated Oct 2026, 10min
Which rail should your agent pay on?
A practical comparison of the four rails AI agents transact on in 2026: cards, bank transfers, stablecoins and x402, and API-native billing. Where each wins, where each fails, and why the choice should be per transaction.
EngineeringAug 2026, 7min
How to set spending limits on AI agents
A practical guide to hard spending limits for AI agents: the four limit layers, approval thresholds, rail-aware enforcement, and the calibration loop that tightens risk without killing throughput.
EngineeringAug 2026, 6min
The agent spending security checklist
A production security checklist for AI agent payments: funding isolation, intent gating, credential handling, tool scoping, hard controls, kill switches, evidence logs, and the drills to run before launch.
The plain truth,
in your inbox.
Occasional, considered notes
on agentic payments. No noise.