What can the AI assistant's own computer do, like running code, working with my files or GitHub?

The AI assistant has its own computer, a sandbox server. It can install packages and run code in pretty much any language. There are built-in runtimes for Python, bash, R, Java and TypeScript, and it can read and write code from GitHub, installing Git if it needs to. You can tell it to install a package and build an integration, and it works out the code. This is available in the AI assistant, in workflows and in web apps.

You can attach files, code and spreadsheets for it to work with. For example, give it a spreadsheet and have it build a dashboard or an application from the data. It can also access your Library files: you can create your own folders for it to work in and store context there, and it can get images and details from your brand kit. You can download the CSV files and slide decks it creates.

For data parsing, ask the AI to write code. Code is more deterministic than instructions, because the same code runs each time.

The computer is mainly for running shell commands and similar tasks. It is not a browser that logs in to other sites for you. If you need an AI with browser access, use an external AI that has it, such as ChatGPT or Claude, and connect it to Votel through MCP.

The assistant also has a small IDE and terminal, so terminal access to the server is available, which is more technical. The easiest way is to tell the AI what you want. Running commands or code on this computer is billed as code execution.