Databao docs

Use Databao UI

Try an end-to-end conversational analytics flow on your own data in Databao UI

In addition to running Databao in the terminal, you can launch it as a Streamlit data app and interact with your data directly from your browser.

Before you begin
  1. Make sure you have access to an LLM:

    • Obtain an API key from your LLM provider, or
    • Install and run a local model using Ollama
  2. If you plan to connect to remote databases, ensure they allow connections from your machine.

Install Databao

If you haven't already, install Databao:

uv add databao
pip install databao

Launch the web interface

In your project directory, run:

databao app

Databao will open a browser tab at http://localhost:8501.

To use a different port, specify it with --server.port, such as databao app --server.port 8502.

Initialize the project

To create a new project in the directory where you launched the app, click Initialize project.

Configure data sources

In the Data sources section, configure at least one data source:

  1. Select the database type, provide credentials, and other connection details.

  2. Click Add data source.

For more information about connecting to databases, see Data sources.

Configure Agent

Before you can run the Databao Agent, you need to select the execution engine and configure the LLM connection.

  1. In Execution engine, select the executor that's most suitable for your use case

  2. In LLM, select the provider and model you want to use. If you are using a cloud LLM, provide the API key.

    Databao supports OpenAI and Anthropic cloud models, other models via OpenAI-compatible APIs, and local models through Ollama.

(Optional) Build context

To improve Databao’s answer accuracy, you can build context by indexing the structure of your data.

To start, click Build context.

Depending on the size and complexity of your data sources, this process may take some time.

Start a new chat

To start a new chat with your data, click New chat. To create more chats, click New chat in the sidebar on the left.

Chats are stored in your project directory, and persist across sessions. To open a previous chat, select it in the sidebar.

Next steps

Now that you've tried building a full conversational analytics flow, you can continue developing your Databao project and expand how the agent understands and queries your data.

  • Add more data sources so that the agent can access more context and provide more accurate answers.

  • Integrate Databao in your workflow by using the open-source Context Engine and Data Agent.

  • Use the Databao platform to bring agentic analytics to your organization at scale. Get in touch with us to launch a proof of concept, discuss your needs, and share feedback.

On this page