At @AIAgentsSummit Berlin, developers explored what happens when AI agents can do more than generate outputs. Using Circle Agent Stack, builders showed agents that could hold wallets, manage budgets, discover paid services, pay for APIs or data, coordinate workflows, and return receipts. Agents are starting to look like economic participants. They need: → Wallets → Budgets → Low-cost payments → Programmable settlement → Receipts → Infrastructure for repeatable machine-to-machine activity That is the kind of future Arc is being built to support.

At @AIAgentsSummit Berlin, developers built with Circle Agent Stack and explored what the agentic economy can look like in practice. We saw agents that could hold wallets, manage budgets, discover paid services, pay in USDC, and return receipts as part of the work they performed. → 25 unique submissions → Agent Wallet + x402 + Gateway payments → Real experiments in agentic commerce Here are the top projects that stood out: 🏆 Proprietor [Winning Project] An agent-run business using Circle Agent Wallet, Circle CLI, and x402 payments to earn revenue, pay supplier agents, manage margin, and return receipts. A great example of what autonomous companies could look like when agents control wallets. 🏆 BytomicProxy A pay-per-request proxy using the Circle Agent Stack starter kit, Agent Wallet, x402/Gateway payments, and spend ledgers to make APIs and infrastructure agent-payable. 🏆 402Cards An agentic commerce adapter using Circle Agent Wallet, ecosystem starter-kit components, and x402 payments to let agents buy items on mainnet through an agent-readable workflow. 🏆 giftr A local gift marketplace using Circle Agent Wallet, Circle Gateway, Arc Testnet, and x402 batching so agents can browse, choose, and pay for gifts with USDC. 🏆 SparkLead A recurring competitor-intelligence workflow using the Circle starter kit, Agent Wallet, and x402 payments so an agent can pay per insight and deliver research. The pattern was clear: The next wave of agents will not just generate answers. They will discover services, pay for what they need, coordinate with other agents, and return receipts. That is the agentic economy Circle Agent Stack is helping builders bring to life.

#Monorepos# make it easier to work across a #codebase#, but they don’t isolate projects the way separate repositories do. #AI# coding agents raise the stakes because they can have broad, persistent access. #GenAI# #AIAgents#

AI models shift in months, but the chips they run on are locked in years earlier. Researchers say #AI# itself can shrink that cycle to weeks by running the entire design flow. #AI# #AIAgents# #GenAI# #GenerativeAI#

Google's answer to recursive self-improvement is to let #agents# "dream" their way through exploration loops, learning from earlier attempts to guide future searches. #AI# #GenAI# #AIAgents# #GenerativeAI#

Most chains treat an AI agent like a script with a hot wallet. Model breaks the moment software starts paying, trading and collaborating without a human at the keyboard. SVP Chain is built for this. #SVPChain# #AIAgents# #AILayer# #Web3#

If your #security# architecture does not evolve alongside the #MCP# update, you will inherit attack surfaces your network-layer tools can't see. #Cybersecurity# #AI# #GenAI# #AIAgents#

#DeepSeek# just raised V4 #API# prices by up to 1,100%—a sharp reminder that “cheap inference” is a moving target. For #AI# builders, model routing, caching and workload scheduling are becoming infrastructure essentials. #GenAI# #AIAgents# #GenerativeAI#

Nobody wants to deal with gas fees, vault rules, or collateral ratios. They just want the end result. ​At the @AIAgentsSummit, @vandynathan and @Zyfai_ CEO @goatv_bk broke down why moving from static vaults to autonomous agents is so critical. and taking all that complexity off the user's hands is how onchain finance actually scales.