In this project, I architected and developed a voice-based automation system, using Make.com’s powerful low-code workflow engine to streamline API communication, data routing, and logic execution. Voice inputs were processed through the VAPI framework, which handled transcription, intent recognition, and multi-agent orchestration.

Main Features

A standout feature of the system was the use of multiple distinct voices to represent different agents, making interactions feel more personalized and intuitive. The agents were interconnected and each of them had specific prompt how to handle intent and redirection.

For example, a friendly voice was being the initial agent who was trying to either answer simple questions or direct user to another “department”, another voice was handling more sensitive questions like answering HR problems and for example what is the procedure to call sick, while a more casual tone handled answering FAQs from the company’s internal wiki.

The final fourth agent was capable of handling complex tasks such as booking meeting rooms, sending notifications to participants, and managing schedules. Upon confirmation of a booking, the assistant would automatically email relevant people with details, ensuring smooth communication and minimizing manual labor.

Other important features

• The system was protected against misuse, I implemented access control mechanisms, ensuring that only authorized uses could interact with the assistant.
• The VAPI platform allowed us to create modular, intelligent agents and these agents collaborated via shared memory and conversational context stored in a vector database, enabling the assistant to carry meaningful conversations over time.
• We also hosted a custom Python scripts used for advanced logging and learning how to apply the intent logic, dynamic prompt construction, and secure webhook handling.

The final result offering a hands-free, intelligent, and secure way to complete tasks and access services. By combining multi-agent orchestration, voice personalization, and automation with strong security, the project delivered a smart, user-friendly voice interface tailored to modern productivity needs.