Why AI Agents Still Won't Pay for Anything — Jan Curn, Apify

Highlights

• Apify operates a marketplace of 80,000 AI tools called actors, paying creators almost $2 million in a single month.

• Gartner predicts that autonomous agents will drive an economy worth $15 trillion by 2028, handling 80% of B2B activity.

• The two leading agentic payment standards, Coinbase's X402 and Stripe's MPP, recorded low transaction volumes of $1 million and $1,500 respectively over a 30-day period.

• Agentic payment protocols primarily compete with API keys and credit cards rather than traditional merchant acquirers.

• Cryptocurrency infrastructure avoids chargebacks entirely, supporting irreversible transactions essential for automated agent interactions.

Timeline

1. 00:00:00 - 07:16:00

   Rise of Agentic Payment Standards

   The rapid expansion of agentic tools created a demand for autonomous financial capabilities. Financial institutions and tech platforms introduced multiple protocols including MasterCard AgentPay, Coinbase X402, Google AP2, and Stripe MPP. Traditional payment rails remain poorly suited for high-frequency microtransactions and chargeback abuse inherent to automated agents, favoring crypto-based alternatives.

   • Apify supports a two-sided marketplace where creators publish and sell web automation tools called actors.
   • Numerous agent payment protocols emerged across fintech and crypto sectors over a 18-month period.
   • Cryptocurrency is positioned to capture agentic payments due to low transaction costs and the elimination of chargebacks.
2. 07:16:00 - 13:31:00

   The Adoption Gap and Funnel Friction

   On-chain monitoring reveals minimal transaction volume for protocols like X402 and MPP, indicating a major gap between protocol availability and actual usage. Developers prefer simple API keys and recurring credit card subscriptions over setting up decentralized wallets for agents. Furthermore, heavy subsidies on AI model tokens discourage agents from purchasing specialized external services when they can synthesize alternatives locally.

   • Real-world transaction volumes for major agent payment standards remain extremely low despite extensive protocol development.
   • Agents heavily utilize commercial data services through standard API keys instead of autonomous payment discovery.
   • Token subsidies offered by major LLM providers reduce economic incentives for agents to purchase external services.
3. 13:31:00 - 20:30:00

   Trust and Payment Architecture Solutions

   Completing transactions with unknown entities requires institutional trust signals or curated directories similar to existing platform app stores. Early payment protocols restricted transactions to fixed pricing, failing to support variable compute costs. Recent updates like X402 batch settlements and MPP sessions introduce escrow-based microtransactions, laying the groundwork for an autonomous agent economy.

   • Autonomous agents require trusted directories and marketplaces to verify service quality before transferring funds.
   • Fixed-price API models are insufficient for complex agent workflows that demand variable pricing structures.
   • Batch settlements and session escrow mechanisms enable secure microtransactions and pay-per-use billing on blockchain networks.