In this informative video, the presenter showcases Verbi, an open-source voice assistant capable of taking actions on behalf of users through function calling and tool usage. Over 14 minutes and 35 seconds, viewers learn about Verbi’s capabilities, setup instructions, and how it operates using state-of-the-art models. The video begins with an engaging demonstration of Verbi checking the user’s calendar and outstanding tasks, highlighting its practical applications for everyday users. The presenter explains how Verbi processes user queries by first converting speech to text, then utilizing a large language model (LLM) to determine if external tools or APIs are needed to generate a response. The assistant’s modular design allows for integration with various models and APIs, making it versatile for different tasks. Detailed instructions on cloning the Verbi repository from GitHub and setting up the necessary environment are provided, along with information on supported actions like retrieving calendar events and checking weather information. The video concludes with a demonstration of Verbi’s capabilities in planning a weekend party, showcasing its ability to pull information from multiple sources and manage tasks effectively.

Prompt Engineering
Not Applicable
September 12, 2024
PT14M35S