Bybit Launches Trading API to Connect to ChatGPT, Claude, and Co
Set Trending Topics as a preferred source on Google.
The race to bring AI-driven trading to European investors has another entrant. Bybit EU, the Vienna-based European arm of crypto exchange Bybit, has just launched API trading. The interface lets developers connect their own trading systems, automated strategies and third-party applications directly to the platform’s infrastructure. On top of that, an open-source component called the “Bybit EU AI Trading Skill” allows AI assistants such as ChatGPT, Claude, Gemini and Codex to link up with a Bybit EU account.
Bybit EU has been licensed in Austria under the EU’s Markets in Crypto-Assets regulation (MiCAR) since last year. The authorisation by the Austrian financial regulator came a few months after the billion-dollar hack of the global exchange. The company has since built out Vienna as the base for its regulated European business and recently added an e-money licence there as well.
What the Interface Can Do
According to the company, the API gives users direct access to market data, portfolio data and order execution. Once the AI connection is set up, users can query prices and holdings, manage spot orders, rebalance portfolios and access spot margin with leverage of up to 10x, all by chatting with their preferred assistant.
“Chat-based AI services have become the preferred way for many people to access information, get things done and interact with digital products. We expect that trend to extend into financial services,” says Georg Harer, Co-CEO of Bybit EU. AI lowers the barrier to building trading software, he adds, while giving experienced developers more efficient ways to scale their solutions.
The launch comes with a 30-day campaign in which Bybit EU is distributing up to 100,000 USDC to eligible users who trade through its API infrastructure. Rewards are volume-based and tied to the applicable terms and conditions, with registration handled through a dedicated landing page.
Skill Instead of MCP Server: The Technical Difference
The timing stands out, because two other providers with Austrian ties have made strikingly similar announcements in recent weeks. Bitpanda opened its multi-asset platform to AI agents, and Scalable Capital let Claude, ChatGPT and Grok reach into customer portfolios. Both rely on the Model Context Protocol (MCP). Bybit EU has taken a different route.
With MCP, the provider runs a server of its own, in Bitpanda’s case at mcp.public.bitpanda.com. The AI assistant connects to that server as a client and is handed a fixed set of tools, eleven of them at Bitpanda: portfolio queries, watchlists, staking and order execution. That gives the provider central control over which functions a model can see at all, how often it may call them and where a user confirmation is mandatory. Bitpanda, for instance, requires a price quote before every order that the user has to confirm explicitly, while Scalable Capital ties every trade and every savings plan to a manual approval.
A skill works differently. It is a package of instructions and scripts that the AI assistant loads and executes in its own environment. The assistant talks to the exchange’s REST interface directly and signs its requests with API keys held locally by the user. There is no provider-operated server in between. That makes the approach more flexible, since the feature set evolves along with the API rather than being limited to a curated toolbox. It also shifts responsibility: key management, permission scoping and safeguards against unintended orders sit more squarely with the user and the assistant. It requires an environment in which the model can execute code.

