On Thursday, AX1 unveiled version 2.0 of AX1 Console, its launchpad on Coinbase's Base network where anyone can put an AI agent to work onchain and compete for weekly prizes, according to a company announcement carried by USA Newshour. AX1 Console 2.0 strips out the setup: pick a token, and Console creates the agent, covers its first runs, and puts it to work. The product is live at console.ax1.vc.

The entry quest, called Verdict, shows how the loop works. According to the announcement, the agent pays for the run through x402, digs through a token's onchain data and market signals, and comes back with a verdict card worth sharing. Every run scores points on a weekly leaderboard, and top players split a prize pool paid in tokenized SpaceX stock (SPCXc) on Base.

Agents as the new paying customers

The announcement's most telling line is about who the customer is. "The internet is getting a new kind of customer," the company said: AI agents that pay for data and services on their own, one request at a time, through x402, the open payment standard introduced by Coinbase. No accounts, no API keys — just USDC, per request. AX1 Console is pitched as the place where that economy stops being a demo and becomes something anyone can run.

The numbers suggest it is not a demo anymore. Since paid quests went live in September, AX1 Console has processed more than 430,000 x402 payments from close to 5,000 agent wallets. Over the last 30 days it has ranked in the top 5 on x402scan, the public x402 explorer, by both transactions and paying agents, and it is listed in the x402 Bazaar on the Coinbase Developer Platform. That makes it one of the first measurable proofs that agents are spending real money onchain at scale.

The agent-as-payer pattern keeps accelerating elsewhere. Mastercard's Agent Pay for Machines program is building card rails for machine-to-machine payments, and the industry is converging on small, per-request payments rather than subscription seats for autonomous software.

Work, proof, and trust

AX1 frames its ecosystem around three things every economy needs: work, proof, and trust, built as one layer in the open. AX1 Console is where agents work. AX1 Assay is how their work gets proven. Partner apps bring the jobs.

AX1 Console's quest and campaign partners already include sport.fun, Upshot, Clash of Coins, and BlockRun. Next on the roadmap, the company said, are partner quests from more apps across Base and head-to-head agent competitions — moving from solo leaderboard grinding toward adversarial, spectator-friendly formats where agents compete directly against each other.

Why AX1 Console matters for the agent economy

Console 2.0 lands at a moment when the agent economy is splitting in two directions. On one side, major platforms are choosing to block agents outright rather than build for them: Amazon blocked Meta's Muse agent, JPMorgan Chase and Charles Schwab bar agents from touching customer accounts, and Cloudflare has blocked agents by default on ad-supported sites since September, as the AI News Brief summarized this week. On the other side, infrastructure like x402, agent payment rails, and now work-marketplaces like Console are being built natively for agents as first-class economic actors.

The bet AX1 Console is making is that the agents-as-customers pattern wins by being useful. An agent that can pay per request needs no enterprise sales cycle, no API key provisioning, and no human in the loop to swipe a card. If AX1 Console's trajectory — from zero to hundreds of thousands of agent payments in weeks — holds, "no-code agent deployment with a prize pool" may turn out to be the on-ramp that makes onchain agent labor a normal thing.

The infrastructure layer is maturing in parallel. Anthropic and OpenAI have donated their Agent Skills standard to the Linux Foundation's Agentic AI Foundation, giving the agent economy neutral-governance plumbing while startups like AX1 build the marketplaces on top.

Sources

AX1 launches Console 2.0, an agentic activities launchpad where AI agents go to work — USA Newshour (via GlobeNewswire); AI News Brief, October 23, 2026