# Overview

### Python SDK

Client library for the [InteractiveAI](https://pypi.org/project/interactiveai/) observability platform.

#### Synopsis

Provides OpenTelemetry-based tracing, automatic LLM framework instrumentation, prompt management, dataset evaluation, and direct API access to InteractiveAI.

### Install

```bash
pip install interactiveai
```

With extras for LLM framework auto-instrumentation:

```bash
pip install "interactiveai[openai]"
pip install "interactiveai[langchain]"
```

#### Quick Start

```python
import interactiveai

client = interactiveai.Interactive(
    public_key="<INTERACTIVEAI_PUBLIC_KEY>",
    secret_key="<INTERACTIVEAI_SECRET_KEY>",
)

with client.start_as_current_span(name="my-operation") as span:
    result = do_work()
    span.update(output=result)

client.flush()
```

#### Options

| Variable                        | Default                      | Description             |
| ------------------------------- | ---------------------------- | ----------------------- |
| `INTERACTIVEAI_PUBLIC_KEY`      | —                            | API public key          |
| `INTERACTIVEAI_SECRET_KEY`      | —                            | API secret key          |
| `INTERACTIVEAI_HOST`            | `https://app.interactive.ai` | API endpoint            |
| `INTERACTIVEAI_DEBUG`           | `false`                      | Enable debug logging    |
| `INTERACTIVEAI_TRACING_ENABLED` | `true`                       | Enable/disable tracing  |
| `INTERACTIVEAI_SAMPLE_RATE`     | `1.0`                        | Sampling rate (0.0–1.0) |