Warning
This repository is archived. These examples use the legacy LaunchDarkly AI SDK for Python
(launchdarkly-server-sdk-ai), which is in maintenance mode and no longer receives new features.
For new applications, use the current Python AI SDK and its examples. To move existing code, see Migrate from the legacy AI SDKs.
| Package | PyPI | Docs |
|---|---|---|
| launchdarkly-server-sdk-ai |  |
Reference |
| launchdarkly-server-sdk-ai-openai |  |
Reference |
| launchdarkly-server-sdk-ai-langchain |  |
Reference |
| launchdarkly-observability |  |
Reference |
Each example is a self-contained application you can run independently to explore LaunchDarkly's AI APIs hands-on. Pick one that matches your provider or use case, follow the README, and you'll be up and running in minutes.
For more comprehensive instructions, visit the Quickstart page or the legacy Python reference guide.
These examples show how to integrate LaunchDarkly AI with different providers.
| Provider | Example | Description |
|---|---|---|
| Bedrock | Converse | completion_config with AWS Bedrock Converse API, metrics tracking |
| Gemini | Generate Content | completion_config with Google GenAI, metrics tracking |
| LangChain | Invoke | completion_config with LangChain, async metrics tracking |
| LangGraph | ReAct Agent | agent_config with a single LangGraph ReAct agent, tool calling, metrics tracking |
| LangGraph | StateGraph | agent_config with multiple LangGraph agents, custom StateGraph workflow, per-node metrics |
| OpenAI | Chat Completions | completion_config with OpenAI, automatic metrics tracking |
These examples demonstrate LaunchDarkly's managed APIs and standalone capabilities.
| Example | Description |
|---|---|
| create_judge | Standalone evaluation of AI responses |
| create_agent | Tool calling, automatic metrics tracking, and judge evaluation |
| create_agent_graph | Multi-node workflows, tool calling, per-node metrics, and judge evaluation |
| create_model | Managed chat, automatic metrics tracking, and judge evaluation |