Conva AI Python Library

This library enables interaction with Conva AI Co-pilots.

Installation

pip install conva-ai

Usage

Initializing the Client

To use the Conva AI library, you first need to initialize the AsyncConvaAI client with your credentials:

from conva_ai import AsyncConvaAI

client = AsyncConvaAI(
    assistant_id="<YOUR_ASSISTANT_ID>",
    assistant_version="<YOUR_ASSISTANT_VERSION>",
    api_key="<YOUR_API_KEY>"
)

Replace the placeholders with your actual assistant ID, version, and API key provided by Conva AI.

Basic Response Generation

To generate a response from the AI, use the invoke_capability method:

import asyncio

query = "What's the weather like today?"
response = asyncio.run(client.invoke_capability(query, stream=False))
print(response.message)

This will send your query to the AI and return a response. The stream=False parameter indicates that you want to receive the full response at once.

Invoking Specific Capabilities

If you want to use a particular capability of the AI, you can specify it using the invoke_capability_name method:

query = "Take me to the home page"
response = asyncio.run(client.invoke_capability_name(query, stream=False, capability_name="navigation"))
print(response.message)

In this example, we're specifically invoking the "navigation" capability of the AI.

Streaming Responses

For longer responses or real-time interactions, you can use streaming mode:

query = "Give me a detailed explanation of quantum computing"
response = asyncio.run(client.invoke_capability(query, stream=True))
async for res in response:
    print(res)

This will print parts of the response as they are generated, allowing for more interactive experiences.

Maintaining Conversation History

To have a continuous conversation with Conva AI, you can keep track of the conversation history:

history = "{}"
while True:
    query = input("Enter your query: ")
    response = asyncio.run(client.invoke_capability(query, stream=False, history=history))
    history = response.conversation_history
    print(response.message)

This loop allows you to have an ongoing conversation with the Conva AI, with each response considering the context of previous interactions.

Debugging Responses

To understand the Conva AI's reasoning behind a response, you can access additional information:

final_response_dict = response.model_dump()
print(final_response_dict["reason"])

This will print out the AI's reasoning process, which can be helpful for debugging or understanding complex responses.

Try It Out

You can experiment with the Conva AI Co-pilot in a Google Colab notebook. This is a great way to get familiar with the library's capabilities without setting up a local environment. Try Conva AI on Google Colab

Additional Information

Make sure to handle API keys securely and never expose them in your code repositories. The library uses asynchronous programming with asyncio for efficient handling of I/O operations.