Picture an AI agent doing your shopping. What does it buy? Almost everyone pictures the same things — it orders dinner, books a flight, restocks the pantry from Amazon. That picture is wrong, or at least badly out of order. Those are some of the last things an agent will buy unattended — with no human tapping "confirm" — not the first.
The word doing the work there is unattended. Plenty of agents already help you shop today; almost none of them complete the purchase without a person in the loop. The first market where an agent can run the whole transaction alone is digital goods — and the cleanest consumer example in that market is a video game key. This isn't a claim about what's cool. It's a claim about which transactions can actually close, start to finish, with no human present.
What has to be true for an agent to buy something alone
Strip a purchase down to mechanics and ask which steps secretly assume a human is sitting there. For an agent to complete a transaction unattended, five things have to hold:
- Settlement is instant on both sides. The agent pays in seconds and receives the goods in seconds. No "ships in 3–5 days," no pending capture, no fulfillment step where a person has to do something.
- No physical logistics. No shipping address, no delivery window, no "sorry you weren't home," no returns that involve a box and a courier. An agent has no hands.
- No per-purchase human tap. No 3-D Secure popup, no "approve this in your banking app," no checkout screen waiting on a thumb. This is the step every consumer shopping agent still keeps — and the one that makes a purchase attended rather than autonomous.
- Low blast radius. A single purchase is small enough ($5–$60, not $5,000) that a mistake is cheap. You can let an agent learn to spend when the worst case is a wasted few dollars, not a wasted flight.
- Agent-verifiable. The agent can confirm, on its own, that it got what it paid for — a deterministic delivery and a payment trail it can check, not a receipt a human eyeballs.
Now run the things people imagine agents buying through that list.
| Criterion | Pizza | Flight | Amazon order | Game key |
|---|---|---|---|---|
| Instant settlement | ~ | ✗ | ✗ | ✓ |
| No physical logistics | ✗ | ~ | ✗ | ✓ |
| No per-purchase human tap | ~ | ✗ | ~ | ✓ |
| Low blast radius | ✓ | ✗ | ~ | ✓ |
| Agent-verifiable | ✗ | ~ | ✗ | ✓ |
✓ clears it cleanly · ~ partial / conditional · ✗ needs a human or a physical step
"But agents already buy physical stuff" — yes, with a human tap
The obvious objection is that this is already happening. It is — and it proves the point rather than breaking it. In late 2025 Amazon shipped "Buy for Me," where its assistant completes purchases on third-party sites; Visa and Mastercard announced agent-payment programs; Google published a protocol for agent payments whose flagship example is buying event tickets the moment they drop. Physical and high-ticket commerce got the agentic spotlight first.
But look at how each one is wired. Above a price threshold they pause for a human to approve. They lean on card-network rails that are now scrambling to bolt on agent authentication. They ship as limited betas with a confirmation step in the middle. The industry is, in effect, conceding the thesis: when the purchase is physical or expensive, you keep a human in the loop. The autonomy the demos advertise is mostly assisted checkout, not unattended spend.
None of these gaps are unsolvable forever. They're just not the first dominoes to fall for true autonomy. The first dominoes are the purchases where none of the human-shaped steps exist at all.
Why digital goods clear the bar — and already do
A digital good has no physical body, so the entire logistics column disappears. Delivery is a string. The agent pays, and a moment later it holds the thing — there is no warehouse, courier, or person standing between payment and possession.
That single property cascades. No delivery means no address to verify. A low ticket means no risk-driven step-up. Instant delivery means the agent can verify the outcome immediately instead of waiting days to find a problem. The whole purchase collapses into something that fits inside an API call and an onchain payment — which is to say, inside the two things an agent actually has.
And this isn't theoretical. The one place fully unattended agent spend already works at scale is crypto rails carrying digital goods: agents paying each other for API calls, inference, compute, and data feeds, settling almost entirely in USDC with no human anywhere in the loop. The machines-paying-machines economy is already live and growing — it just happens to be invisible because nobody's buying anything a person would recognize. Yet.
Why game keys specifically
Let's be precise, because this is where the thesis earns or loses credibility. The first digital goods agents bought autonomously were inputs for the agents themselves — API access, compute, data. That's real, it's the bulk of unattended agent spend today, and any honest version of this argument has to start there. Gift cards and top-ups are a close consumer-side peer, and others are already wiring those onto agent rails too.
So game keys aren't "the first thing an agent ever buys." They're something more specific and, we'd argue, more interesting: the cleanest consumer-intent purchase an agent can complete end to end. "Buy this game for my friend" is a job people already hand to assistants, and it maps onto an autonomous agent without a single human-shaped step. A few reasons it's the cleanest of the consumer options:
- The catalog is liquid and deep. More than 40,000 distinct SKUs across every platform and region. "Find the best option" is a real reasoning task, not a lookup — which is exactly the kind of thing you'd want an agent for. A gift card, by contrast, is fungible at face value: there's nothing to reason over.
- Price is transparent and comparable. The same title trades across many marketplaces, so "is this a good price?" has an answer the agent can compute and act on. Most digital goods have no liquid reference price; keys do.
- The demand is consumer-pull. Gifting, rewards, giveaways, "buy this for me." These are jobs already delegated to assistants, which is why the use cases aren't hypothetical.
- It's a one-shot transaction. No subscription to manage, no recurring billing relationship, no reorder logistics. The agent buys, delivers, done — the simplest possible commerce primitive.
For the record, the keys themselves are sourced from authorized wholesale distributors, not the grey market — the kind of provenance question an agent (or its owner) should be able to take for granted.
What was missing, and why it works now
If digital goods are such an obvious fit, why hasn't this existed? Two pieces were missing, and both arrived recently.
The first is payment built for machines. Card rails assume forms, billing addresses, and human-approved challenges. x402 moves payment into HTTP itself: the server answers with status 402 and structured instructions, the agent pays onchain in USDC, the server verifies the transaction and returns the goods. No accounts, no callbacks, no human tap. A payment an agent can complete and a merchant can verify — both without a person.
The second is an interface an agent can read. A catalog built for humans — paginated marketing pages, geo-IP pricing, copy stuffed into titles — forces an agent to scrape and guess. An agent-native API and an MCP server give it structured filters, one product per request, and competitor prices right in the response, so it can decide instead of parse. (That last part — a seller shipping its rivals' prices inside its own API — sounds backwards until you realize an agent-native merchant has the opposite incentive of a human storefront: the agent will check competitors anyway, so you win by saving it the round trip.)
Neither piece is large. Together they turn "buy me that game" from a wall an agent bounces off into three HTTP calls and one wallet signature.
This isn't a forecast — but it isn't mainstream yet either
The point of this post isn't to predict that agents will shop someday. The capability for this narrow slice exists right now. But be honest about the scale: adoption is early, and the loudest agentic-commerce launches are still human-in-the-loop. That contrast is the actual story — mainstream consumer agents are stuck waiting on a human tap, while the digital, crypto-native path is quietly already autonomous.
That's the slice we built rails for. If you want to see what the agent code actually looks like, the companion post walks through building a game-buying agent in about 200 lines with x402 + MCP. Live API and OpenAPI spec: api.cdk.bot/docs. Other agents can discover us programmatically at /.well-known/agent.json.
Have a use case we haven't thought of, or a product category you wish an agent could buy? info@cdk.bot.