> For the complete documentation index, see [llms.txt](https://aita-1.gitbook.io/aita/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://aita-1.gitbook.io/aita/reference/agent.md).

# Aita agent

Aita agent is an AI agent that runs on LLM (Large Language Model), and can be used to ask questions in natural language. Aita agent can also use tools (refer to [Aita tool](https://github.com/project-aita/aita/blob/main/docs/reference/agent/tools.md)) to let agent to run on various data APIs such as Pandas, Pyarrow, Spark, Pytorch etc.

## Models

Aita model providers are inspired by [jupyter-ai-magic](https://jupyter-ai.readthedocs.io/en/latest/users/index.html#model-providers), and currently supported models are:

| Provider               | Provider ID            | Environment variable(s)    | Python package(s)                         |
| ---------------------- | ---------------------- | -------------------------- | ----------------------------------------- |
| AI21                   | `ai21`                 | `AI21_API_KEY`             | `ai21`                                    |
| Anthropic              | `anthropic`            | `ANTHROPIC_API_KEY`        | `langchain-anthropic`                     |
| Anthropic (playground) | `anthropic-playground` | `ANTHROPIC_API_KEY`        | `langchain-anthropic`                     |
| Bedrock                | `bedrock`              | N/A                        | `boto3`                                   |
| Bedrock (playground)   | `bedrock-playground`   | N/A                        | `boto3`                                   |
| Cohere                 | `cohere`               | `COHERE_API_KEY`           | `cohere`                                  |
| ERNIE-Bot              | `qianfan`              | `QIANFAN_AK`, `QIANFAN_SK` | `qianfan`                                 |
| Gemini                 | `gemini`               | `GOOGLE_API_KEY`           | `langchain-google-genai`                  |
| GPT4All                | `gpt4all`              | N/A                        | `gpt4all`                                 |
| Hugging Face Hub       | `huggingface_hub`      | `HUGGINGFACEHUB_API_TOKEN` | `huggingface_hub`, `ipywidgets`, `pillow` |
| NVIDIA                 | `nvidia-playground`    | `NVIDIA_API_KEY`           | `langchain_nvidia_ai_endpoints`           |
| OpenAI                 | `openai`               | `OPENAI_API_KEY`           | `langchain-openai`                        |
| OpenAI (playground)    | `openai-playground`    | `OPENAI_API_KEY`           | `langchain-openai`                        |
| SageMaker              | `sagemaker-endpoint`   | N/A                        | `boto3`                                   |

## Example

{% code title="python" %}

```python
from aita.agent.pandas import PandasAgent
import pandas as pd

df = pd.DataFrame({
    'customer_id': [1, 2, 3, 4, 5],
    'purchase_amount': [100, 200, 300, 400, 500]
})
sql_agent = PandasAgent(df, "gpt-3.5-turbo")

print(sql_agent.stream("I want to get the top customers which making the most purchases"))
```

{% endcode %}

## Agents

Currently supported agents are:

* [PythonAgent](/aita/reference/agent/python.md)
* [SqlAgent](/aita/reference/agent/sql.md)
* [PandasAgent](/aita/reference/agent/pandas.md)
* [PyarrowAgent](/aita/reference/agent/pyarrow.md)
* [SparkAgent](/aita/reference/agent/spark.md)