> ## Documentation Index
> Fetch the complete documentation index at: https://ottic.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Chat & the sandbox

> The chat workspace where you delegate, and the cloud sandbox that lets the agent actually do the work.

The chat workspace is where you delegate, and the cloud sandbox is what lets the agent actually do the work. Together they are the core of Ottic.

## The chat workspace

You work with your agent in a streaming chat. As it works, you see everything happen live:

* **Activity stream** — the tools it calls and the steps it takes, in real time.
* **Reasoning** — the thinking behind a decision, shown as it works through a problem.
* **Artifacts** — finished outputs (reports, exports, documents) appear as cards you can open.
* **File upload** — drop in a spreadsheet, PDF, or document and ask the agent to work with it.

Nothing is a black box. You can watch what the agent did and why, and step in whenever you want.

## The cloud sandbox

Behind the chat, your agent runs in an isolated cloud sandbox with a real terminal and a real browser. That is what separates Ottic from tools that only talk:

* **Run code** — Python and shell for data processing, calculations, file manipulation, and chart generation.
* **Drive a real browser** — Playwright for live web research, navigating sites, scraping content, and taking screenshots.
* **Read and write files** — produce CSVs, documents, and reports you can download.

So when you ask for a competitive teardown, it can actually open the competitor's site. When you ask to reconcile a spreadsheet, it can actually run the code. The work is done, not just described.

## Isolated per user

Each user's agent runs in its own isolated sandbox, so your work, files, and credentials stay separate from everyone else's.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.