# Pull payments, credit, and schlep blindness 8/29/26 Most people have the intuitive direction of payments wrong. Remember: your landlord *pulls* your rent payment, Amex *pulls* your card payment, the IRS *pulls* your tax balance. You're not "sending" them these payments – they're *debiting* you for goods and services rendered. A "payment" is, ideally, a motion organized by the end-recipient. The end-recipient acts, the sender merely *authorizes.* This truth is the origin of a lot of intermediation. The first paragraph of the Bitcoin whitepaper – a screed against intermediation – is pointed at the anodyne but fundamental payments problem of "disputes." Disputes are a symptom of merchants being able to optimistically *pull* funds from customers, and customers therefore being able to dispute them. Bitcoin is literally — explicitly — a rebellion against pull payments. ![[Screenshot 2026-05-19 at 12.11.25 PM.png|The opening paragraph of the Bitcoin whitepaper.|700]] It's easy to characterize the intermediated status quo as rent-seeking. Or, more intelligently, as symptomatic of a pre-technological era. But the case being made is less frequently "the thing these middlemen do could be done for cheaper," and more often "must we bother with the things that they do?" I particularly liked this observation:  This is a form of [schlep blindness](https://www.paulgraham.com/schlep.html) (h/t Paul Graham). > The most dangerous thing about our dislike of schleps is that much of it is unconscious. Your unconscious won't even let you see ideas that involve painful schleps. That's schlep blindness. PG's evergreen post was explicitly an ode to Stripe; it points to payments as the classic "schlep." Still, *even* payments people – even at companies like these, even *I*! – often forget the *schleps* that underpin what "payments" *actually are.* ![[Pasted image 20260501170219.png|Early schlep noticers.|700]] If you had to *actively push* funds to a merchant every time you owed them for a good, service, or subscription: it would be enough work for you that at the margins – perhaps even *at* *large* – far fewer of these transactions would occur in the first place. It's instructive to walk all the way back to paper checks and work forward into "debiting". ## Paper checks were the future Sending someone money once required a physical transfer (the sort of thing Satoshi waxes nostalgically about in his whitepaper). Checks were a step *forward* from that status quo. They're a simple, flexible piece of paper that the sender can offer the recipient – which the recipient can show to their bank as proof that they're authorized to *pull* the sender's funds. When you scan a paper check using your mobile banking app: you are *debiting* – pulling from – the person who gave you that check. Remember, payments are a motion organized by the end-recipient: ie. _pulled_. ## Crypto I'm actually going to talk about crypto *before* we get into ACH – which runs a highly-functional debiting rail. 16 years after Bitcoin, stablecoins — USDC and USDT — have taken payments by storm. Satoshi, at least as a technologist, was right! They’re useful means by which to store US dollars and transmit them globally, without the travails of legacy FX systems, hard-to-obtain multi-currency bank accounts, or slow SWIFT. Stablecoin payments are still blockchain-based, irrevocable\*, push payments — in the evolutionary path of Bitcoin. But their utility to the mainstream financial system comes with a cost to the vision in Satoshi’s first paragraph: intermediation — *engaged* intermediation —is in *major* demand. In April of 2026, the Drift Protocol on Solana was compromised, customer funds being funneled out using USDC as a medium (to North Korea, as it happens). The entire industry wanted Circle — the issuer of USDC, holder of the actual dollars and treasury backing the token 1:1 — to *intervene.*  Circle didn’t do anything to stop it, much to the chagrin of many industry participants. Tether, issuer of rival stablecoin USDT, who works with law enforcement far more than you’d expect if you had a cursory familiarity with their operation (famously more of a black box than Circle), used their willingness to freeze funds as a marketing opportunity.  People want their money intermediated; it's a lot of responsibility to be your own bank. It is a real advancement that you can *choose* to be your own – crypto still produces a genuinely powerful "public option." The team at [Better Money Company](https://bettermoney.com) is building the first clearinghouse for stablecoins. They [observed](https://bettermoney.com/insights/the-drift-hack-another-reason-to-have-your-own-stablecoin) the actual insight buried in the news. > If a business is building on somebody else's stablecoin, they are at the whim of that stablecoin issuer’s procedures, preferences, and incentives. Drift presumably attempted to get the USDC frozen over those ~6 critical hours...unsuccessfully. Consumers have a far worse chance of recovering their funds as a result. **If a business launches its own stablecoin, it controls the big red button.** If stablecoins are to mature as a payment rail – you need a few things to happen at once. There is a baseline level of "responsibility" that stablecoin issuers like Circle, Tether, and others have taken on in the background – around regulatory & compliance, treasury management, etc – that people do not fully appreciate. Stablecoins aren't purely regulatory arbitrage; they're quite regulated, and in some ways *favorable* to regulators (in their utter transparency and [[The great irony of stablecoin|centralization]]). But you need the institutions serving actual end-customers – the fintechs, "stablecoin neobanks," etc – to take on more of the actual *responsibilities* of an issuer; and therefore the controls and powers. Especially since mistakes are almost as irrevocable as they are with physical cash. You need them to become the issuers in this equation: or at least you need more issuers. Simultaneously – these new-issuances still have to clear amongst themselves, 1:1, so that they remain fungible enough for the whole thing to remain viable. This is where a clearinghouse enters the picture. In a world where there are many institutions issuing their own stablecoins, and they're reliably clearing between each other: you can extrapolate the frontier from other clearing systems. It's a long road there. ## ACH and cards ACH is the gold standard for payroll, statement payments, B2B, rent, deposits, and almost everything else. It's cheaper than wires, as a consequence of being batched + netted (banks only exchange the net inflow/outflow of their batch, rather than each individual transaction). The value-prop of "netting" has also made its way into the crypto discourse:  But to be honest, netting is besides the point. What matters is that you can use ACH to *pull,* cheaply. ACH debit (ie. pull) is how — again — "Amex pulls your statement payment, your landlord pulls your rent," and so on. They're able to pull from you, optimistically & without your explicit confirmation each time, because they've collected your "authorization," are subject to penalties if they pull from you otherwise, and *you* have a significant degree of recourse — months — to go to your bank and dispute the pull to get your funds back from them. There is also an invisible extension of credit involved in the whole mechanism. If you were to (rightfully) dispute that pull, but the funds had already been spent on the other side (whether withdrawn or paid out): the account on the other side would start to have a negative balance. In other words: * The account-holder would *owe* their bank those funds back * By letting the account-holder spend those funds: the bank was extending them *credit* This is how card pulls work, too: a merchant who pulls from their end-customer is similarly exposed to the risk that their customer initiates and wins a "chargeback" against them: leaving them short of the funds (whether or not the funds were already withdrawn/spent). The bank is similarly extending credit to any merchant it allows to accept card payments. And credit requires a **trusted** relationship: built on KYC/B, and on either an underwriting of the account-holder (ie. the merchant) or a tolerance for some threshold of losses. As a bank, you need a means of recoupment and collections, and reason to believe you won't simply lose the money. The intermediaries in payments — banks, issuers & acquirers, networks — thereby make it possible, in their extension of credit, and in their (significant) work on risk — for the end-recipients of payments to *pull* from their customers. To offer some contrast: in *push*-heavy payment workflows — eg. Accounts Payable (the bills a company has to pay, often 30 days after services are received) — the customer is often *pushing* an ACH or wire to their vendor. But this means the *vendor* is having to underwrite + extend credit to their customer: constraining how many customers they can support, at which scale. By taking this work off of the merchant or vendor's plate, and applying economies of scale: the intermediary who enables pull payments expands the radius of customers who the merchant can transact with, and is therefore expanding the degree to which commerce occurs at large. --- Things have evolved (I started writing this in April 2026, stalled, and am closing it out in August). Stablecoins — the "push" payment rail this post was intended to implicitly challenge — are already ceding more ground to cards in "agentic payments." Instinct and Grok Bot, two popular AI assistants, announced core integrations with Stripe Link: effectively a wallet for your existing cards (think: when you type in your phone number at checkout, and your card is automatically filled out). These two agents, for the time being, will use your existing cards as payment methods wherever checkout is powered by Stripe. *Pull* is already starting to win in this nascent category. Stripe, extending "credit" to as many merchants as it does, can offer pull via Link / Cards in a way that it can't via even its own stablecoin offerings. People viscerally presume that they have recourse, via chargeback, in the event that their AI assistant makes a purchase that goes against their intent. At [Natural](https://natural.com), we're extending "credit" — implicit and [explicit](https://www.natural.com/blog/100m-credit-facility) — for agentic payments. We're doing this both in the old-world manner that Stripe is successfully modeling (with cards), and in a more agent-first manner. We (necessarily) let your agents accept card payments or "pay by bank" — ie. pull payments. Notably even over the phone, powered by voice agents. But in going post-card, post-checkout, we're trending closed-loop: both merchants and customers live on the same network, designed for their respective agents: the agents facilitate transactions *between* them, in the course of accomplishing higher-level objectives. "Closed-loop payments" reduce to ledger transfers, comparable on their face to eg. stablecoins: but even they ultimately succeed on the basis of us extending merchants credit and underwriting them, in order for these merchants to exist on this network, be trustworthy. Critically: for customers to have recourse in the form of disputes, and for these disputes to be "fair" enough to merchants in a world with agents running around, such that merchants are incentivized to _accept_ payments from agents. The "cold-start" of this marketplace kicks off in earnest by building _trust_ between buyers and sellers: agents are a trust _reset_. With trust re-established, agents can take this work over in the course of their greater goals. Lots of schleps left to [solve](https://www.natural.com/careers)!