Quant Intelligence · Desktop

Your first session (about 20 minutes)

This walkthrough matches the download email: install the desktop app, paste your 14-day Basic key, add one LLM API key, create the bundled conda environment, refresh built-in MCP libraries, then run one live-data prompt (Futu + Python) end to end.

Before you open the app

Request the installer from the product page — the email includes macOS / Windows / Linux links and your trial license key.
Install Miniconda or Anaconda if you do not have it yet. The app bundles environment.yml (same file as in the atabet-quant-intelligence repo) to create one env named atabet_env for Data, Trader, Terminal, and local Python tools.
Have at least one LLM provider API key ready (Gemini, OpenAI, Claude, DeepSeek, etc.). Keys stay on your machine only.
Optional for the sample prompt: install Futu OpenD, log in, and keep it running so the agent can pull Hong Kong quotes through the Futu channel.

Get the installer Watch the 2‑minute demo

Install and launch

Open the installer from your download email, drag Atabet Quant Intelligence into Applications (macOS) or run the setup wizard (Windows). On first launch, paste the license key from the same email when the app asks for it.

1
Open Settings
Click the gear icon in the header. You will return here for API keys, Python, and license renewal.
2
Add an LLM API key
Open Models & keys, choose a provider, and save an API key locally. You do not paste Atabet Data or Trader tokens here — those are injected from your license when the app launches. The 14-day Basic trial starts at first successful activation (see Account & app), not when the email was sent.
3
Set up local Python
Open Local runtime. Once, run conda env create -f environment.yml using the copy inside the app bundle (macOS: right-click the app → Show Package Contents → Resources; Windows/Linux: alongside the executable). In development, use the repo environment.yml at the project root. Select Conda: atabet_env and click Test execution. Atabet Data/Trader access tokens are filled in automatically from your license. Only add optional third-party keys (e.g. Tushare) if you need those channels.
4
Refresh MCP libraries
Click Add MCP Tools in the sidebar. For built-in packages such as atabet_data and atabet_trader, choose Latest (or Use latest) and click Apply on each row so the agent gets up-to-date API docs and samples. Applying can also install that version into your selected Python env. Click Done when finished.
5
Open a project folder (optional)
Use the folder icon in the sidebar to open a workspace (for example a new empty folder). The agent can read and write files there, run Python, and save charts or notes. You can skip this for the sample chat-only prompt below.
6
Send one live-data prompt
Pick a model that has an API key configured. With Futu OpenD running, try: Can we use futu to download data and build a chart to compare the performance of Tencent and HSI index? The agent should call the Futu channel through atabet_data and run Python locally.
7
Follow Python output in chat
Tool cards show Python execution, stdout, and errors inline. Approve package installs or shell commands when prompted. When the run succeeds, tables and Plotly charts can appear below the tool card. If Futu is offline, fix OpenD first and retry the same prompt.