aipotheosis 0.0.1b3

The official Python SDK for the Aipolabs (Aipotheosis Labs) API

Aipolabs Python SDK

The official Python SDK for the Aipolabs API. Currently in private beta, breaking changes are expected.

The Aipolabs Python SDK provides convenient access to the Aipolabs REST API from any Python 3.10+ application.

Documentation

The REST API documentation is available here.

Installation

pip install aipolabs

or with poetry:

poetry add aipolabs

Usage

Aipolabs platform is built with agent-first principles. Although you can call each of the APIs below any way you prefer in your application, we strongly recommend trying the Agent-centric features and taking a look at the examples to get the most out of the platform and to enable the full potential and vision of future agentic applications.

Client

from aipolabs import Aipolabs

client = Aipolabs(
    # it reads from environment variable by default so you can omit it if you set it in your environment
    api_key=os.environ.get("AIPOLABS_API_KEY")
)

Apps

Types

from aipolabs.types.apps import App, AppDetails

Methods

# search for apps, returns list of basic app data, sorted by relevance to the intent
# all parameters are optional
apps: list[App] = client.apps.search(intent="I want to search the web", limit=10, offset=0)

# get detailed information about an app, including functions supported by the app
app_details: AppDetails = client.apps.get(app_name="BRAVE_SEARCH")

Functions

Types

from aipolabs.types.functions import Function, FunctionExecutionResult

Methods

# search for functions, returns list of basic function data, sorted by relevance to the intent
# all parameters are optional
functions: list[Function] = client.functions.search(
    app_names=["BRAVE_SEARCH", "TAVILY"],
    intent="I want to search the web",
    limit=10,
    offset=0
)

# get function definition of a specific function, this is the schema you can feed into LLM
# the actual format is defined by the inference provider, currently only openai is supported
function_definition: dict = client.functions.get(function_name="BRAVE_SEARCH__WEB_SEARCH")

# execute a function with the provided parameters
result: FunctionExecutionResult = client.functions.execute(
    function_name="BRAVE_SEARCH__WEB_SEARCH",
    function_parameters={"query": "what is the weather in barcelona"}
)

if result.success:
    print(result.data)
else:
    print(result.error)

Agent-centric features

The SDK provides a suite of features and helper functions to make it easier and more seamless to use functions in LLM powered agentic applications. This is our vision and the recommended way of trying out the SDK.

Meta Functions and Unified Function Calling Handler

We provide 4 meta functions that can be used with LLMs as tools directly, and a unified handler for function calls. With these the LLM can discover apps and functions (that our platform supports) and execute them autonomously.

from aipolabs import meta_functions

# meta functions
tools = [
    meta_functions.AipolabsSearchApps.SCHEMA,
    meta_functions.AipolabsSearchFunctions.SCHEMA,
    meta_functions.AipolabsGetFunctionDefinition.SCHEMA,
    meta_functions.AipolabsExecuteFunction.SCHEMA,
]

# unified function calling handler
result = client.handle_function_call(
    tool_call.function.name, json.loads(tool_call.function.arguments)
)

There are mainly two ways to use the platform with the meta functions:

• Fully Autonomous: Provide all 4 meta functions to the LLM, please see the agent_with_dynamic_function_discovery_and_fixed_tools.py for more details.
• Semi Autonomous: Provide all but AipolabsExecuteFunction to the LLM, and use the Unified Function Calling Handler to execute functions, please see the agent_with_dynamic_function_discovery_and_dynamic_tools.py for more details.

Please also see agent_with_preplanned_tools.py for comparison where the specific functions are pre selected and provided to the LLM.