Circle Launches Arc Studio, an AI Coding Agent for Building Onchain Applications
Key Takeaways
- •Arc Studio generates full-stack blockchain applications from plain-language instructions, including frontend code, backend logic, smart contracts and tests.
- •The tool can deploy applications to Arc and several other EVM-compatible testnets while working alongside established coding tools such as Claude Code, Codex and Cursor.
- •Circle launched Studio as part of a broader stack combining Arc, USDC as native gas, Arc App Kits and Agent Stack infrastructure for AI agents.
- •Circle warns that AI-generated responses may be inaccurate and that developers should review and test smart contracts before deployment.
- •The tool may broaden access to blockchain development, with adoption among Arc’s more than 100 ecosystem participants serving as a near-term signal.

Circle has launched Arc Studio, an AI coding agent that enables developers to build onchain applications, smart contracts and AI agents using natural-language prompts. The company introduced the tool on Sept. 17, one day after launching the public mainnet of its Arc blockchain.
Arc Studio is designed to reduce the technical effort required to move an onchain idea from concept to working application. According to Circle, it can generate frontend interfaces, backend logic, smart contracts, tests and deployment workflows within a single development process. The release arrives as Circle expands Arc beyond core blockchain infrastructure into a broader platform for payments, financial applications and agent-driven activity.
From prompts to working applications
Developers can describe an application in plain language and use Arc Studio to generate a full-stack implementation. Unlike a general-purpose coding assistant, the tool is built specifically around Circle's blockchain infrastructure.
Its key capabilities include:
- Generating smart contracts and application code from natural-language instructions
- Building frontend interfaces around existing contracts
- Creating applications that leverage Circle's stablecoin infrastructure
- Testing and reviewing generated contracts before deployment
- Deploying applications to Arc and several other EVM-compatible testnets
Circle says Studio can also work alongside existing developer tools, including Claude Code, Codex and Cursor, allowing developers to adopt the agent without overhauling their established coding workflows. The compatibility places the launch within a broader industry shift, as AI coding assistants move prompt-based programming into everyday use across software development.
AI and stablecoins converge
Arc Studio forms part of Circle's broader strategy to connect stablecoins with AI-driven economic activity. The Arc mainnet launched with USDC as its native gas currency, together with more than 100 applications and institutional or ecosystem builders. Circle has also introduced Arc App Kits, which provide developers with tools for payments, swaps, onramps and yield. Separately, its Agent Stack gives AI agents wallets and infrastructure for holding and moving assets under predefined policies.
Read together, the pieces form a vertically connected stack: Arc provides the chain, USDC the native gas, App Kits the payment building blocks and Agent Stack the agent infrastructure, with Studio now serving as the entry point for assembling applications on top of them.
The company views these products as infrastructure for an emerging economy in which software agents can execute financial tasks without constant human intervention. That said, AI-generated smart contracts still require careful testing and review. Circle's own Studio interface warns developers that generated responses can be inaccurate and that code should be reviewed before use.
As Arc develops, Studio could give Circle a direct way to attract more developers while making blockchain application development accessible beyond specialist smart-contract engineers. Near-term signals include how quickly builders among Arc's 100-plus ecosystem participants adopt the tool and whether deployment support expands beyond Arc and the EVM-compatible testnets currently supported.