Generative AIDemo not connected
Multimodal AI Assistant Platform
An assistant that combines text, voice and images, orchestrating tool calls against a SQLite-backed flight workflow.
- Python
- Gemini API
- Tool Calling
- Whisper
- SQLite
- Gradio
Problem
Explore how a single assistant can understand natural-language requests, take actions through tools, and respond across text, voice and image.
Approach
Exposed SQLite-backed functions to the Gemini API as callable tools so the model could orchestrate multiple, dependent tool calls in one turn. Added Whisper transcription, image generation and text-to-speech in a Gradio Blocks interface. The flight search and booking flow is a project workflow for demonstrating tool use, not a commercial service.
Key features
- 01Function / tool calling with multiple dependent calls per turn
- 02SQLite persistence for the sample flight workflow
- 03Session-based access control for verified users
- 04Voice input via Whisper transcription and spoken responses via text-to-speech
- 05Image generation inside the same conversation
Demo
A working deployment hasn't been connected for this project yet. It will appear here once a hosted link is added.
Next project
Multi-Model Code Converter