# FAQ > Short answers about what Claude can and cannot do in your Orbital project: trades, projects, data, teammates, apps, ChatGPT, advice and disconnecting. Source: https://orbitalaip.com/docs/claude-and-chatgpt/faq > [!IMPORTANT] Disclaimer > Backtest results are hypothetical and do not predict future performance. Nothing here is investment advice: Orbital computes the rules you supplied and recommends nothing. Orbital is not a registered investment adviser or broker-dealer. Costs are as modelled by the chosen cost profile and may differ from live trading. ## Does Claude place trades? No. The Orbital AI connector has no tool that places an order, deploys a strategy or reaches Trade, so trading stays inside Orbital. See [What Claude can and cannot do](https://orbitalaip.com/docs/claude-and-chatgpt/connecting#what-claude-can-and-cannot-do). ## Which project does Claude see? Exactly one: the project you chose on the **Connect to Orbital** page. Claude cannot reach any other project or workspace, even ones you belong to; see [One workspace, one project](https://orbitalaip.com/docs/claude-and-chatgpt/connecting#one-workspace-one-project). ## Can I switch to another workspace or project? Yes, by connecting again: remove the connection, add it again, and pick the other workspace and project on the consent page, whose pickers start on the workspace and project you are using in Orbital at that moment. Orbital has no page for creating a second project today, so in practice this means another workspace; see [One workspace, one project](https://orbitalaip.com/docs/claude-and-chatgpt/connecting#one-workspace-one-project). ## Does Claude see my positions, account or other apps? No: the tools reach only the connected project's strategies and backtests, with their versions, runs, results and trade logs; nothing in Trade, Watchlists, Chat or Agents has a tool, and no tool lists or runs your other workflows. Claude can also read your account's email address, when the connection includes "See your email address", and learn what a run costs in credits and, when a run is refused for credits, how many remain; see [What Claude can and cannot do](https://orbitalaip.com/docs/claude-and-chatgpt/connecting#what-claude-can-and-cannot-do). ## What does Orbital keep from the conversation? Only what a tool call carries: the arguments Claude sends, such as strategy rules, program source or backtest settings. Orbital keeps the strategies, versions and runs those calls create, as if you had made them in the app. The rest of your conversation never reaches Orbital, and what Claude keeps is governed by Anthropic's own policies. Orbital's [privacy policy](https://orbitalaip.com/legal/privacy-policy) applies to what Orbital keeps. ## Where are the strategies and results? In Orbital, in the connected project. Open **Backtesting** in the sidebar: the **Strategies** page lists the project's strategies, those Claude made included. Open a strategy for its **Overview**; once it has a run, it also shows **Report**, **Trades** and **Runs**. Every run Claude starts appears under **Runs**, and Claude reads the same records you see there. ## Can I keep editing in Orbital what Claude made? Yes: a strategy Claude creates is an ordinary strategy in your project, and you edit it like any other. After you save in Orbital, Claude must read the strategy again before its next edit, so nothing is overwritten. Claude can also edit a draft strategy you made, unless it is deployed, automated or run by another workflow. Tested in Backtesting is not a lock: a strategy with runs stays editable. While a deployment runs, Claude's edit of that strategy is refused with a message to branch it in Orbital; see [What a running deployment locks](https://orbitalaip.com/docs/backtesting/manage-and-delete-strategies#what-a-running-deployment-locks). To branch it, open the strategy's **Workflow** (or **Code** for a Program), clone a version with **Versions**, the version's three-dot menu and **Clone**, and ask Claude to edit the clone, as [Restore and clone](https://orbitalaip.com/docs/workflows/runs-and-versions#restore-and-clone) describes. ## Can teammates connect to the same project? Yes, if they already belong to the workspace and can open the project in Orbital. Each teammate connects from their own Claude, signed in to their own Orbital account, and everyone connected sees the same strategies and runs. Each connection carries that person's role, so a viewer can read but cannot change strategies or run backtests; see [the role refusal](https://orbitalaip.com/docs/claude-and-chatgpt/troubleshooting#your-role-in-this-project-cannot-change-strategies-or-spend-credits). ## Does it work in the Claude mobile and desktop apps? Orbital AI is a custom connector on your Claude account, added in Claude on the web or in the desktop app. Which Claude apps can use custom connectors is set by Claude, so check Claude's documentation for the desktop and mobile apps. Claude Code in the terminal connects with its own command; [Connect Claude to Orbital](https://orbitalaip.com/docs/claude-and-chatgpt/connecting) has the steps for both. ## Is ChatGPT supported? Not yet. Orbital AI for ChatGPT is not yet tested or listed, so Orbital supports nothing in ChatGPT today; until it ships, Claude and Claude Code are the assistants that connect to Orbital. See [ChatGPT](https://orbitalaip.com/docs/claude-and-chatgpt/chatgpt). ## Are the results advice? No. Orbital computes the rules you supplied on historical data and recommends nothing, and every backtest result Claude reads carries the notice that opens this page. See [What Claude can and cannot do](https://orbitalaip.com/docs/claude-and-chatgpt/connecting#what-claude-can-and-cannot-do). ## How do I disconnect? Remove the **Orbital AI** connector in Claude, or the server in Claude Code, as [Removing the connection](https://orbitalaip.com/docs/claude-and-chatgpt/connecting#removing-the-connection) describes.