ZeroClick launched on Product Hunt with a one-line pitch: “Sell your product to AI agents.” The product page at zeroclick.ai is more specific. ZeroClick turns a product into “agent-purchasable x402 & MPP services, discoverable on all major indexes.” The dashboard on that same page shows 620 agent transactions and $1.9K in revenue.
Read those two numbers together. If they are real and current, the average agent transaction on ZeroClick is roughly three dollars. That is not a business yet. It is a signal about where the business is forming.
The mechanism matters more than the metrics. x402 is the HTTP 402 “Payment Required” status code revived as a machine-payable rail, and MPP points at the same idea from a different direction: an endpoint that an autonomous agent can call, get quoted a price for, and pay without a human in the loop. ZeroClick’s job is to wrap an existing product as one of those endpoints and then get it listed wherever agents look for services. The screenshot on the homepage shows the flow in miniature: 8f2c → dataco.dev /enrich/batch sits in a “Waiting for human” state, while 3b91 → lexi.ai /v1/translate, c04a → pixelforge.ai /generate/image, and 77d5 → relaymail.io /send are all “Approved.”
The approval step is the tell
Look at what ZeroClick is actually selling. Four of the five sample transactions are API-shaped: batch enrichment, translation, image generation, email send. These are the products that were already machine-callable. ZeroClick is not teaching agents to buy things they could not buy before. It is adding a discovery layer and a payment handshake on top of services that already existed as HTTP endpoints.
That is a real gap. An agent that wants to translate a document has no reliable way to find lexi.ai /v1/translate, compare it against a dozen competitors, and settle payment in the same request. Today that work falls to a human writing a client, managing an API key, and reconciling an invoice. ZeroClick’s bet is that the discovery-plus-settlement layer becomes the choke point, and whoever owns it collects a toll.
The counterargument is that x402 and MPP are open rails. If the protocol is standard, the discovery layer is a commodity, and the margin collapses to whatever the indexes charge for placement. ZeroClick’s own homepage hints at this: “discoverable on all major indexes” is a distribution promise, not a moat. Anyone can list on the same indexes.
The $1.9K question
The revenue figure deserves scrutiny. $1.9K is not disclosed as daily, weekly, or lifetime. A dashboard that shows a running transaction counter and a running revenue counter, with no time axis, is a marketing artifact as much as an operational one. Tessera has not verified whether the figure is cumulative since launch or a trailing window, and ZeroClick does not say on the page. Treat it as directional.
Directional still tells you something. Six hundred twenty transactions is enough to prove the rail works end to end. It is nowhere near enough to prove demand. For comparison, the API marketplaces ZeroClick is implicitly competing with, from RapidAPI to the cloud vendors’ own marketplaces, move orders of magnitude more volume, and none of them have solved agent-native payments either.
The more interesting number is the one ZeroClick is not showing: how many of those 620 transactions were initiated by an autonomous agent versus a human testing the flow. The “Waiting for human” state in the sample is a reminder that most of this stack still routes through a person at the decision point. An agent that can pay but still needs approval is a checkout form with extra steps.
The second ZeroClick
There is a naming collision worth flagging, because it muddies the picture. A separate operation at zeroclicklabs.ai calls itself an “AI digital marketing agency” and pitches “AI SEO” services: optimization for Google AI Overviews, ChatGPT, Gemini, Perplexity, and “voice search.” It claims 200+ satisfied clients and cites a statistic that “as many as 43% people use AI tools daily to search for products and services.” Its testimonials are first-name-only and read like template copy. The two ZeroClicks are not obviously the same company, and the marketing agency’s claims are unverified.
The collision is not incidental. Both are selling the same underlying anxiety: that AI intermediaries now sit between a product and its buyer, and that whoever controls the intermediary’s view of your product controls the sale. One sells you a storefront for agents. The other sells you visibility inside the models. Both are early, both are small, and both are betting that the human click is going away.
An agent that can pay but still needs approval is a checkout form with extra steps.
What this means for builders
If you ship an API, the practical question ZeroClick raises is not whether to list on it. It is whether your product is legible to a machine buyer at all. Can an agent discover your endpoint, read its price, understand its input and output schema, and pay for a single call without a signup flow? Most products cannot answer yes to all four. The ones that can are the ones that will show up in agent indexes first, and the ones that cannot will keep paying for human attention while the agent traffic routes around them.
The compute economics cut against the hype in the near term. Agent-initiated transactions are cheap by construction, which means the take rate has to be either very high or the volume very large before the model works. $1.9K of revenue at a few dollars per call is a rounding error against the infrastructure cost of running the rail. ZeroClick’s real product may end up being the index listing and the compliance layer around agent payments, not the transactions themselves.
Watch the time axis. If ZeroClick starts publishing transactions per day, and that number climbs past the tens of thousands, the agent-commerce thesis has legs. If the dashboard keeps showing a cumulative counter with no window, the pitch is still ahead of the demand. The 620 number is the honest one on the page. Everything else is positioning.