AI Voice Powered Restaurant Assistant

Agentic Foodie is a voice-powered restaurant ordering app built using Streamlit for the frontend UI. It features a modern interface for browsing the menu, recording voice input, and managing orders. Key Streamlit components like st.columns, st.button, and st.session_state handle layout and interaction, while custom CSS enhances styling and responsiveness. audio_recorder_streamlit enables in-browser audio recording, returning WAV bytes. These are processed with pydub (AudioSegment) and speech_recognition (Recognizer with recognize_google) to convert audio to text for voice-based interactions. Groq API (LLaMA 3) provides the core NLP via HTTP requests made using the requests library. The app sends user messages and history to the API and receives structured JSON responses with intents, item IDs, quantities, and replies. Rate-limiting is managed using x-ratelimit-remaining-requests and timestamps with the time module. JSON is used throughout for defining menus, promotions, and exchanging data with the Groq API. The io module (io.BytesIO) handles in-memory audio processing for seamless transcription. A custom Multi-agent Control Plane (MCP) organizes tasks like checking availability and suggesting items. Agents are registered using decorators (e.g., @inventory_agent.task) and simulate modular multi-agent interactions. Overall, this tech stack enables an AI-driven, interactive restaurant ordering experience with support for voice input, menu browsing, upselling, and conversational responses — all in a sleek, real-time Streamlit app.