The Latitude Python SDK provides a convenient way to interact with the Latitude platform from your Python applications.

​
Installation

The Latitude SDK is compatible with Python 3.9 or higher.

pip install latitude-sdk
# or
poetry add latitude-sdk
# or
uv add latitude-sdk

​
Authentication and Initialization

Import the SDK and initialize it with your API key. You can generate API keys in your Latitude project settings under “API Access”.

import os
from latitude_sdk import Latitude

latitude = Latitude(os.getenv("LATITUDE_API_KEY"))

You can also provide additional options during initialization:

latitude = Latitude(os.getenv("LATITUDE_API_KEY"), LatitudeOptions(
    project_id=12345, # Your Latitude project ID
    version_uuid="optional-version-uuid", # Optional version UUID
))

Keep your API key secure and avoid committing it directly into your codebase.

​
Examples

Check out our cookbook for more examples of how to use the Latitude SDK.

​
SDK Usage

The Latitude Python SDK is an async library by design. This means you must use it within an async event loop, such as FastAPI or Async Django. Another option is to use the built-in asyncio library.

import asyncio
from latitude_sdk import Latitude

latitude = Latitude("your-api-key-here")

async def main():
    prompt = await latitude.prompts.get("prompt-path")
    print(prompt)

asyncio.run(main())

​
SDK Structure

The Latitude SDK is organized into several namespaces:

  • prompts: Methods for managing and running prompts
  • logs: Methods for creating and managing logs
  • evaluations: Methods for triggering evaluations and creating results

​
Prompt Management

​
Get a Prompt

To retrieve a specific prompt by its path:

prompt = await latitude.prompts.get('prompt-path')

​
Get All Prompts

To retrieve all prompts in your project:

prompts = await latitude.prompts.get_all()

​
Get or Create a Prompt

To get an existing prompt or create a new one if it doesn’t exist:

prompt = await latitude.prompts.get_or_create('prompt-path')

You can also provide the content when creating a new prompt:

prompt = await latitude.prompts.get_or_create('prompt-path', GetOrCreatePromptOptions(
    prompt='This is the content of my new prompt',
))

​
Running Prompts

​
Non-Streaming Run

Execute a prompt and get the complete response once generation is finished:

result = await latitude.prompts.run('prompt-path', RunPromptOptions(
    parameters={
        'productName': 'CloudSync Pro',
        'audience': 'Small Business Owners',
    },
    # Optional: Provide a custom identifier for this run
    custom_identifier='email-campaign-2023',
    # Optional: Provide callbacks for events
    on_finished=lambda result: print('Run completed:', result.uuid),
    on_error=lambda error: print('Run error:', error.message),
))

print('Conversation UUID:', result.uuid)
print('Conversation messages:', result.conversation)

​
Handling Streaming Responses

For real-time applications (like chatbots), use streaming to get response chunks as they are generated:

async def on_event(event: StreamEvent):
    if event.event == StreamEvents.Provider and event.type == 'text-delta':
        print(event.text_delta)
    elif event.event == StreamEvents.Latitude and event.type == ChainEvents.ChainCompleted:
        print('Conversation UUID:', event.uuid)
        print('Conversation messages:', event.messages)

await latitude.prompts.run('prompt-path', RunPromptOptions(
    parameters={
        'productName': 'CloudSync Pro',
        'audience': 'Small Business Owners',
    },
    # Enable streaming
    stream=True,
    # Provide callbacks for events
    on_event=on_event,
    on_finished=lambda result: print('Stream completed:', result.uuid),
    on_error=lambda error: print('Stream error:', error.message),
))

​
Using Tools with Prompts

You can provide tool handlers that the model can call during execution:

async def get_weather(arguments: Dict[str, Any], details: OnToolCallDetails) -> Dict[str, Any]:
    # `arguments` contains the arguments passed by the model
    # `details` contains context like tool id, name, messages...
    # The result can be anything JSON serializable
    return { "weather": "sunny" }

await latitude.prompts.run('prompt-path', RunPromptOptions(
    parameters={
        'query': 'What is the weather in San Francisco?',
    },
    # Define the tools the model can use
    tools={
        'getWeather': get_weather,
    },
))

If you need to pause the execution of a tool, you can do so by returning details.pause_execution() in the tool handler. You can resume the conversation later by returning the tool results in the latitude.prompts.chat method.

​
Chat with a Prompt

Follow the conversation of a runned prompt:

messages = [
    {
        'role': 'user',
        'content': 'Hello, how can you help me today?',
    },
]

result = await latitude.prompts.chat('conversation-uuid', messages, ChatPromptOptions(
    # Chat options are similar to the run method
    on_finished=lambda result: print('Chat completed:', result.uuid),
    on_error=lambda error: print('Chat error:', error.message),
))

print('Conversation UUID:', result.uuid)
print('Conversation messages:', result.conversation)

Messages follow the PromptL format. If you’re using a different method to run your prompts, you’ll need to format your messages accordingly.

​
Rendering Prompts

​
Prompt Rendering

Render a prompt locally without running it:

result = await latitude.prompts.render(
    'Your prompt content here with {{ parameters }}',
    RenderPromptOptions(
        parameters={
            'topic': 'Artificial Intelligence',
            'tone': 'Professional',
        },
        # Optional: Specify a provider adapter
        adapter=Adapter.OpenAI,
))

print('Rendered config:', result.config)
print('Rendered messages:', result.messages)

​
Chain Rendering

Render a chain of prompts locally:

async def on_step(messages: list[MessageLike], config: dict[str, Any]) -> str | MessageLike:
    # Process each step in the chain
    print('Processing step with messages:', messages)
    # Return a string or a message object
    return 'Step response'

result = await latitude.prompts.render_chain(
    Prompt(
        path='prompt-path',
        content='Your prompt content here with {{ parameters }}',
        provider='openai',
    ),
    on_step,
    RenderChainOptions(
        parameters={
            'topic': 'Machine Learning',
            'complexity': 'Advanced',
        },
        # Optional: Specify a provider adapter
        adapter=Adapter.OpenAI,
))

print('Rendered config:', result.config)
print('Rendered messages:', result.messages)

​
Logging

​
Creating Logs

Push a log to Latitude manually for a prompt:

messages = [
    {
        'role': 'user',
        'content': 'Hello, how can you help me today?',
    },
]

log = await latitude.logs.create('prompt-path', messages, CreateLogOptions(
    response='I can help you with anything!',
))

​
Evaluations

​
Triggering Evaluations

Trigger an evaluation manually for a conversation:

result = await latitude.evaluations.trigger('conversation-uuid', TriggerEvaluationOptions(
    # Optional: trigger all or specific evaluations
    evaluation_uuids=['eval-uuid-1', 'eval-uuid-2'],
))

​
Creating Evaluation Results

Push a result to Latitude manually for an evaluation:

result = await latitude.evaluations.create_result('conversation-uuid', 'evaluation-uuid', CreateEvaluationResultOptions(
    # The result can be a string, boolean, or number
    result=True,
    reason='I liked it!',
))

​
Complete Method Reference

​
Initialization

# SDK initialization
class LatitudeOptions:
    promptl: Optional[PromptlOptions]
    internal: Optional[InternalOptions]
    project_id: Optional[int]
    version_uuid: Optional[str]
    tools: Optional[dict[str, OnToolCall]]

Latitude(
    api_key: str,
    options: Optional[LatitudeOptions]
)

​
Prompts Namespace

# Get a prompt
class GetPromptOptions:
    project_id: Optional[int]
    version_uuid: Optional[str]

class GetPromptResult: Prompt

latitude.prompts.get(
    path: str,
    options: Optional[GetPromptOptions]
) -> GetPromptResult:

# Get all prompts
class GetAllPromptsOptions:
    project_id: Optional[int]
    version_uuid: Optional[str]

class GetAllPromptsResult: List[Prompt]

latitude.prompts.get_all(
    options: Optional[GetAllPromptsOptions]
) -> GetAllPromptsResult:

# Get or create a prompt
class GetOrCreatePromptOptions:
    project_id: Optional[int]
    version_uuid: Optional[str]
    prompt: Optional[str]

class GetOrCreatePromptResult: Prompt

latitude.prompts.get_or_create(
    path: str,
    options: Optional[GetOrCreatePromptOptions]
) -> GetOrCreatePromptResult:

# Run a prompt
class RunPromptOptions:
    project_id: Optional[int]
    version_uuid: Optional[str]
    on_event: Optional[OnEvent]
    on_finished: Optional[OnFinished]
    on_error: Optional[OnError]
    custom_identifier: Optional[str]
    parameters: Optional[dict[str, Any]]
    tools: Optional[dict[str, OnToolCall]]
    stream: Optional[bool]

class RunPromptResult:
    uuid: str
    conversation: list[Message]
    response: ChainResponse
    tool_requests: list[ToolCall]

latitude.prompts.run(
    path: str,
    options: Optional[RunPromptOptions]
) -> Optional[RunPromptResult]:

# Chat with a prompt
class ChatPromptOptions:
    on_event: Optional[OnEvent]
    on_finished: Optional[OnFinished]
    on_error: Optional[OnError]
    tools: Optional[dict[str, OnToolCall]]
    stream: Optional[bool]

class ChatPromptResult:
    uuid: str
    conversation: list[Message]
    response: ChainResponse
    tool_requests: list[ToolCall]

latitude.prompts.chat(
    uuid: str,
    messages: Sequence[MessageLike],
    options: Optional[ChatPromptOptions]
) -> Optional[ChatPromptResult]:

# Render a prompt
class RenderPromptOptions:
    parameters: Optional[dict[str, Any]]
    adapter: Optional[Adapter]

class RenderPromptResult:
    messages: list[MessageLike]
    config: dict[str, Any]

latitude.prompts.render(
    prompt: str,
    options: Optional[RenderPromptOptions]
) -> RenderPromptResult:

# Render a chain
class RenderChainOptions:
    parameters: Optional[dict[str, Any]]
    adapter: Optional[Adapter]

class RenderChainResult:
    messages: list[MessageLike]
    config: dict[str, Any]

latitude.prompts.render_chain(
    prompt: Prompt,
    on_step: OnStep,
    options: Optional[RenderChainOptions]
) -> RenderChainResult:

​
Logs Namespace

# Create a log
class CreateLogOptions:
    project_id: Optional[int]
    version_uuid: Optional[str]
    response: Optional[str]

class CreateLogResult: Log

latitude.logs.create(
    path: str,
    messages: Sequence[MessageLike],
    options: Optional[CreateLogOptions]
) -> CreateLogResult:

​
Evaluations Namespace

# Trigger an evaluation
class TriggerEvaluationOptions:
    evaluation_uuids: Optional[List[str]]

class TriggerEvaluationResult:
    evaluations: List[str]

latitude.evaluations.trigger(
    uuid: str,
    options: Optional[TriggerEvaluationOptions]
) -> TriggerEvaluationResult:

# Create an evaluation result
class CreateEvaluationResultOptions:
    result: Union[str, bool, int]
    reason: str

class CreateEvaluationResultResult: EvaluationResult

latitude.evaluations.create_result(
    uuid: str,
    evaluation_uuid: str,
    options: CreateEvaluationResultOptions
) -> CreateEvaluationResultResult:

​
Error Handling

The SDK raises ApiError instances when API requests fail. You can catch and handle these errors:

from latitude_sdk import ApiError

async def handle_errors():
    try:
        prompt = await latitude.prompts.get("non-existent-prompt")
    except ApiError as error:
        print(f"API Error: {error.message}")
        print(f"Error Code: {error.code}")
        print(f"Status: {error.status}")
    except Exception as error:
        print(f"Unexpected error: {error}")

​
Logging Features

  • Automatic Logging: All runs through latitude.prompts.run() are automatically logged in Latitude, capturing inputs, outputs, performance metrics, and trace information.
  • Custom Identifiers: Use the optional custom_identifier parameter to tag runs for easier filtering and analysis in the Latitude dashboard.
  • Response Identification: Each response includes identifying information like uuid that can be used to reference the specific run later.

​
Further Information